{"meta": {"count": 1000, "db_response_time_ms": 44, "page": 1, "per_page": 100, "groups_count": null, "x_query": {"oql": "get works where retracted is (false)\n  and year is (2025)\n  and title-abstract has ((stemmed \"RLHF\" OR stemmed \"RLVR\" OR stemmed \"instruction tuning\" OR stemmed \"low rank adaptation\" OR stemmed \"parameter efficient fine tuning\" OR stemmed \"preference optimization\" OR stemmed \"reinforcement learning from human feedback\" OR stemmed \"supervised fine tuning\") AND (stemmed \"language model\" OR stemmed \"large language model\" OR stemmed \"vision language model\"))\n  and type is (article or preprint or review);\nthen sample (1000) of those works with seed (42)", "oqo": {"get_rows": "works", "filter_rows": [{"column_id": "is_retracted", "value": false}, {"column_id": "publication_year", "value": 2025}, {"join": "or", "filters": [{"column_id": "title_and_abstract.search", "value": "\"RLHF\"", "operator": "has"}, {"column_id": "title_and_abstract.search", "value": "\"RLVR\"", "operator": "has"}, {"column_id": "title_and_abstract.search", "value": "\"instruction tuning\"", "operator": "has"}, {"column_id": "title_and_abstract.search", "value": "\"low rank adaptation\"", "operator": "has"}, {"column_id": "title_and_abstract.search", "value": "\"parameter efficient fine tuning\"", "operator": "has"}, {"column_id": "title_and_abstract.search", "value": "\"preference optimization\"", "operator": "has"}, {"column_id": "title_and_abstract.search", "value": "\"reinforcement learning from human feedback\"", "operator": "has"}, {"column_id": "title_and_abstract.search", "value": "\"supervised fine tuning\"", "operator": "has"}]}, {"join": "or", "filters": [{"column_id": "title_and_abstract.search", "value": "\"language model\"", "operator": "has"}, {"column_id": "title_and_abstract.search", "value": "\"large language model\"", "operator": "has"}, {"column_id": "title_and_abstract.search", "value": "\"vision language model\"", "operator": "has"}]}, {"join": "or", "filters": [{"column_id": "type", "value": "article"}, {"column_id": "type", "value": "preprint"}, {"column_id": "type", "value": "review"}]}], "sample": 1000, "seed": "42"}, "url": "/works?filter=is_retracted:false,publication_year:2025,title_and_abstract.search:%22RLHF%22|%22RLVR%22|%22instruction tuning%22|%22low rank adaptation%22|%22parameter efficient fine tuning%22|%22preference optimization%22|%22reinforcement learning from human feedback%22|%22supervised fine tuning%22,title_and_abstract.search:%22language model%22|%22large language model%22|%22vision language model%22,type:article|preprint|review&sample=1000&seed=42&per_page=100&page=1"}, "cost_usd": 0.001}, "results": [{"id": "https://openalex.org/W4416814762", "doi": "https://doi.org/10.1093/jamiaopen/ooaf157", "title": "Span-based annotation framework for LLM-based clinical named entity recognition: development and validation using Korean emergency department notes", "publication_year": 2025, "publication_date": "2025-11-03", "type": "article", "language": "en", "authorships": [{"author_position": "first", "author": {"id": null, "display_name": "Eun Hye Jang", "orcid": "https://orcid.org/0009-0005-8685-211X", "observed_orcids": ["https://orcid.org/0009-0005-8685-211X"]}, "institutions": [{"id": "https://openalex.org/I848706", "display_name": "Sungkyunkwan University", "ror": "https://ror.org/04q78tk20", "country_code": "KR", "type": "education", "lineage": ["https://openalex.org/I848706"]}], "countries": ["KR"], "is_corresponding": false, "raw_author_name": "Eun Hye Jang", "raw_affiliation_strings": ["Department of Digital Health, SAIHST, Sungkyunkwan University , Seoul, 06355,", "Smart Health Lab, SAIHST, Sungkyunkwan University , Seoul, 06355,"], "raw_orcid": "https://orcid.org/0009-0005-8685-211X", "affiliations": [{"raw_affiliation_string": "Department of Digital Health, SAIHST, Sungkyunkwan University , Seoul, 06355,", "institution_ids": ["https://openalex.org/I848706"]}, {"raw_affiliation_string": "Smart Health Lab, SAIHST, Sungkyunkwan University , Seoul, 06355,", "institution_ids": ["https://openalex.org/I848706"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5069825770", "display_name": "Javier Aguirre", "orcid": "https://orcid.org/0009-0007-6802-944X", "observed_orcids": ["https://orcid.org/0009-0007-6802-944X"]}, "institutions": [{"id": "https://openalex.org/I848706", "display_name": "Sungkyunkwan University", "ror": "https://ror.org/04q78tk20", "country_code": "KR", "type": "education", "lineage": ["https://openalex.org/I848706"]}], "countries": ["KR"], "is_corresponding": false, "raw_author_name": "Javier Aguirre", "raw_affiliation_strings": ["Smart Health Lab, SAIHST, Sungkyunkwan University , Seoul, 06355,"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Smart Health Lab, SAIHST, Sungkyunkwan University , Seoul, 06355,", "institution_ids": ["https://openalex.org/I848706"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5036127587", "display_name": "Sangji Lee", "orcid": "https://orcid.org/0009-0001-8364-3719", "observed_orcids": ["https://orcid.org/0009-0001-8364-3719"]}, "institutions": [{"id": "https://openalex.org/I848706", "display_name": "Sungkyunkwan University", "ror": "https://ror.org/04q78tk20", "country_code": "KR", "type": "education", "lineage": ["https://openalex.org/I848706"]}], "countries": ["KR"], "is_corresponding": false, "raw_author_name": "Sangji Lee", "raw_affiliation_strings": ["Department of Digital Health, SAIHST, Sungkyunkwan University , Seoul, 06355,", "Smart Health Lab, SAIHST, Sungkyunkwan University , Seoul, 06355,"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Department of Digital Health, SAIHST, Sungkyunkwan University , Seoul, 06355,", "institution_ids": ["https://openalex.org/I848706"]}, {"raw_affiliation_string": "Smart Health Lab, SAIHST, Sungkyunkwan University , Seoul, 06355,", "institution_ids": ["https://openalex.org/I848706"]}]}, {"author_position": "middle", "author": {"id": null, "display_name": "Hyeyoon Moon", "orcid": null, "observed_orcids": []}, "institutions": [{"id": "https://openalex.org/I848706", "display_name": "Sungkyunkwan University", "ror": "https://ror.org/04q78tk20", "country_code": "KR", "type": "education", "lineage": ["https://openalex.org/I848706"]}], "countries": ["KR"], "is_corresponding": false, "raw_author_name": "Hyeyoon Moon", "raw_affiliation_strings": ["Department of Digital Health, SAIHST, Sungkyunkwan University , Seoul, 06355,", "Smart Health Lab, SAIHST, Sungkyunkwan University , Seoul, 06355,"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Department of Digital Health, SAIHST, Sungkyunkwan University , Seoul, 06355,", "institution_ids": ["https://openalex.org/I848706"]}, {"raw_affiliation_string": "Smart Health Lab, SAIHST, Sungkyunkwan University , Seoul, 06355,", "institution_ids": ["https://openalex.org/I848706"]}]}, {"author_position": "last", "author": {"id": null, "display_name": "Won Chul Cha", "orcid": null, "observed_orcids": []}, "institutions": [{"id": "https://openalex.org/I2802194831", "display_name": "Samsung Medical Center", "ror": "https://ror.org/05a15z872", "country_code": "KR", "type": "healthcare", "lineage": ["https://openalex.org/I2250650973", "https://openalex.org/I2802194831"]}, {"id": "https://openalex.org/I848706", "display_name": "Sungkyunkwan University", "ror": "https://ror.org/04q78tk20", "country_code": "KR", "type": "education", "lineage": ["https://openalex.org/I848706"]}], "countries": ["KR"], "is_corresponding": true, "raw_author_name": "Won Chul Cha", "raw_affiliation_strings": ["Department of Digital Health, SAIHST, Sungkyunkwan University , Seoul, 06355,", "Department of Emergency Medicine, Samsung Medical Center, Sungkyunkwan University School of Medicine , Seoul,", "Smart Health Lab, SAIHST, Sungkyunkwan University , Seoul, 06355,"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Department of Digital Health, SAIHST, Sungkyunkwan University , Seoul, 06355,", "institution_ids": ["https://openalex.org/I848706"]}, {"raw_affiliation_string": "Department of Emergency Medicine, Samsung Medical Center, Sungkyunkwan University School of Medicine , Seoul,", "institution_ids": ["https://openalex.org/I2802194831", "https://openalex.org/I848706"]}, {"raw_affiliation_string": "Smart Health Lab, SAIHST, Sungkyunkwan University , Seoul, 06355,", "institution_ids": ["https://openalex.org/I848706"]}]}], "abstract_inverted_index": {"Objective:": [0], "This": [1, 193], "study": [2, 194], "aims": [3], "to": [4], "develop": [5], "and": [6, 32, 82, 135, 142, 212, 227, 233], "validate": [7], "of": [8], "a": [9, 58, 101, 151, 196], "span-based": [10, 140, 143, 198], "annotation": [11, 43, 199], "framework": [12, 200], "for": [13, 163, 201, 231], "clinical": [14, 29, 165, 203, 214], "named": [15], "entity": [16, 39, 222], "recognition": [17], "(NER)": [18], "using": [19], "large": [20], "language": [21, 85], "models": [22], "(LLMs)": [23], "based": [24], "on": [25, 67, 92], "Korean": [26, 62], "emergency": [27], "department": [28], "notes.": [30], "Materials": [31], "Methods:": [33], "Two": [34], "datasets": [35], "with": [36, 50, 76, 158], "the": [37, 51, 93, 112, 116, 181], "same": [38], "types": [40], "but": [41], "different": [42], "spans": [44, 223], "(word-": [45], "vs": [46], "phrase-level)": [47], "were": [48, 74], "constructed,": [49], "phrase-level": [52, 95, 228], "dataset": [53, 96], "further": [54, 98], "was": [55, 65, 97], "expanded": [56], "into": [57], "doubled": [59, 94], "version.": [60], "A": [61], "language-specific": [63], "LLM": [64, 81], "fine-tuned": [66, 83, 91], "each": [68], "dataset,": [69], "producing": [70], "three": [71, 109, 207], "variants": [72, 110], "that": [73, 170, 224], "compared": [75], "two": [77], "baseline": [78], "models,": [79], "few-shot": [80, 159], "small": [84], "model": [86], "(SLM).": [87], "The": [88, 123, 177], "final": [89, 124, 178], "variant": [90, 125, 179], "evaluated": [99], "against": [100], "human": [102, 182], "annotator.": [103], "Results:": [104], "In": [105, 210], "all": [106, 121, 132], "experimental": [107], "settings,": [108, 215], "outperformed": [111, 180], "baselines": [113], "by": [114, 206], "achieving": [115], "highest": [117], "F1": [118, 127], "scores": [119, 128], "across": [120, 131], "metrics.": [122], "achieved": [126], "exceeding": [129], "0.80": [130], "averaging": [133], "strategies": [134], "evaluation": [136], "metrics,": [137], "including": [138], "token-based,": [139], "exact,": [141], "partial": [144], "evaluations": [145], "demonstrating": [146], "its": [147, 185], "robustness": [148], "applicable": [149], "in": [150, 219], "practical": [152], "setting.": [153], "Discussion:": [154], "While": [155], "prompt": [156], "engineering": [157], "is": [160, 174], "widely": [161], "adopted": [162], "LLM-based": [164, 202], "NER,": [166], "our": [167], "results": [168], "proved": [169], "supervised": [171], "fine-tuning": [172], "(SFT)": [173], "consistently": [175], "superior.": [176], "annotator,": [183], "emphasizing": [184], "potential": [186], "as": [187], "an": [188], "automatic": [189], "labeling": [190], "tool.": [191], "Conclusion:": [192], "introduced": [195], "novel": [197], "NER": [204], "verified": [205], "independent": [208], "experiments.": [209], "multilingual": [211], "real-world": [213], "LLMs": [216], "have": [217], "proven": [218], "handling": [220], "complex": [221], "include": [225], "word-level": [226], "annotations,": [229], "particularly": [230], "long": [232], "attribute-rich": [234], "entities.": [235]}, "primary_topic": {"id": "https://openalex.org/T11710", "display_name": "Biomedical Text Mining and Ontologies", "score": 0.6689000129699707, "subfield": {"id": "https://openalex.org/subfields/1312", "display_name": "Molecular Biology"}, "field": {"id": "https://openalex.org/fields/13", "display_name": "Biochemistry, Genetics and Molecular Biology"}, "domain": {"id": "https://openalex.org/domains/1", "display_name": "Life Sciences"}}, "topics": [{"id": "https://openalex.org/T11710", "display_name": "Biomedical Text Mining and Ontologies", "score": 0.6689000129699707, "subfield": {"id": "https://openalex.org/subfields/1312", "display_name": "Molecular Biology"}, "field": {"id": "https://openalex.org/fields/13", "display_name": "Biochemistry, Genetics and Molecular Biology"}, "domain": {"id": "https://openalex.org/domains/1", "display_name": "Life Sciences"}}, {"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.2678000032901764, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T13702", "display_name": "Machine Learning in Healthcare", "score": 0.05249999836087227, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "doi:10.1093/jamiaopen/ooaf157", "is_oa": true, "landing_page_url": "https://doi.org/10.1093/jamiaopen/ooaf157", "pdf_url": null, "source": {"id": "https://openalex.org/S4210237468", "display_name": "JAMIA Open", "issn_l": "2574-2531", "issn": ["2574-2531"], "is_oa": true, "is_in_doaj": true, "is_core": true, "listed_in": ["cwts-core", "doaj", "doyens", "ki-jl-1", "norway-1"], "host_organization": "https://openalex.org/P4310311647", "host_organization_name": "University of Oxford", "host_organization_lineage": ["https://openalex.org/P4310311647"], "host_organization_lineage_names": ["University of Oxford"], "type": "journal"}, "license": "cc-by-nc", "license_id": "https://openalex.org/licenses/cc-by-nc", "version": "publishedVersion", "is_accepted": true, "is_published": true, "raw_source_name": "JAMIA Open", "raw_type": "journal-article"}, "cited_by_count": 1, "is_retracted": false}, {"id": "https://openalex.org/W4415070848", "doi": "https://doi.org/10.48550/arxiv.2506.13300", "title": "Seewo's Submission to MLC-SLM: Lessons learned from Speech Reasoning Language Models", "publication_year": 2025, "publication_date": "2025-06-16", "type": "preprint", "language": "en", "authorships": [{"author_position": "first", "author": {"id": "https://openalex.org/A5100758169", "display_name": "Bo Hu Li", "orcid": "https://orcid.org/0000-0001-5980-4861", "observed_orcids": ["https://orcid.org/0000-0001-5980-4861"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Li, Bo", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5034706235", "display_name": "Chengben Xu", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Xu, Chengben", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "last", "author": {"id": "https://openalex.org/A5030057663", "display_name": "Wufeng Zhang", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Zhang, Wufeng", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}], "abstract_inverted_index": {"This": [0, 76], "paper": [1], "presents": [2], "Seewo's": [3], "systems": [4], "for": [5, 44, 51, 98, 106], "both": [6], "tracks": [7], "of": [8, 96, 104, 115], "the": [9, 82, 87, 113], "Multilingual": [10], "Conversational": [11], "Speech": [12], "Language": [13], "Model": [14], "Challenge": [15], "(MLC-SLM),": [16], "addressing": [17], "automatic": [18], "speech": [19, 41], "recognition": [20], "(ASR)": [21], "and": [22, 38, 62, 101], "speaker": [23], "diarization": [24], "with": [25, 65], "ASR": [26], "(SD-ASR).": [27], "We": [28], "introduce": [29], "a": [30, 94, 102], "multi-stage": [31], "training": [32], "pipeline": [33], "that": [34], "explicitly": [35], "enhances": [36], "reasoning": [37], "self-correction": [39, 72], "in": [40], "language": [42], "models": [43], "ASR.": [45], "Our": [46], "approach": [47, 77], "combines": [48], "curriculum": [49], "learning": [50], "progressive": [52], "capability": [53], "acquisition,": [54], "Chain-of-Thought": [55], "data": [56], "augmentation": [57], "to": [58, 69], "foster": [59], "intermediate": [60], "reflection,": [61], "Reinforcement": [63], "Learning": [64], "Verifiable": [66], "Rewards": [67], "(RLVR)": [68], "further": [70], "refine": [71], "through": [73], "reward-driven": [74], "optimization.": [75], "achieves": [78], "substantial": [79], "improvements": [80], "over": [81], "official": [83], "challenge": [84, 119], "baselines.": [85], "On": [86], "evaluation": [88], "set,": [89], "our": [90], "best": [91], "system": [92], "attains": [93], "WER/CER": [95], "11.57%": [97], "Track": [99, 107], "1": [100], "tcpWER/tcpCER": [103], "17.67%": [105], "2.": [108], "Comprehensive": [109], "ablation": [110], "studies": [111], "demonstrate": [112], "effectiveness": [114], "each": [116], "component": [117], "under": [118], "constraints.": [120]}, "primary_topic": {"id": "https://openalex.org/T10201", "display_name": "Speech Recognition and Synthesis", "score": 0.9994999766349792, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T10201", "display_name": "Speech Recognition and Synthesis", "score": 0.9994999766349792, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10860", "display_name": "Speech and Audio Processing", "score": 9.999999747378752e-05, "subfield": {"id": "https://openalex.org/subfields/1711", "display_name": "Signal Processing"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T12031", "display_name": "Speech and dialogue systems", "score": 9.999999747378752e-05, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "pmh:oai:arXiv.org:2506.13300", "is_oa": true, "landing_page_url": "http://arxiv.org/abs/2506.13300", "pdf_url": "https://arxiv.org/pdf/2506.13300", "source": {"id": "https://openalex.org/S4306400194", "display_name": "arXiv (Cornell University)", "issn_l": "2331-8422", "issn": ["2331-8422"], "is_oa": true, "is_in_doaj": false, "is_core": false, "listed_in": [], "host_organization": "https://openalex.org/I205783295", "host_organization_name": "Cornell University", "host_organization_lineage": ["https://openalex.org/I205783295"], "host_organization_lineage_names": [], "type": "repository"}, "license": null, "license_id": null, "version": "submittedVersion", "is_accepted": false, "is_published": false, "raw_source_name": null, "raw_type": "text"}, "cited_by_count": 0, "is_retracted": false}, {"id": "https://openalex.org/W4407424722", "doi": "https://doi.org/10.48550/arxiv.2502.07374", "title": "LLMs Can Easily Learn to Reason from Demonstrations Structure, not content, is what matters!", "publication_year": 2025, "publication_date": "2025-02-11", "type": "preprint", "language": "en", "authorships": [{"author_position": "first", "author": {"id": "https://openalex.org/A5016324529", "display_name": "Dacheng Li", "orcid": "https://orcid.org/0009-0005-1895-1933", "observed_orcids": ["https://orcid.org/0009-0005-1895-1933"]}, "institutions": [{"id": "https://openalex.org/I95457486", "display_name": "University of California, Berkeley", "ror": "https://ror.org/01an7q238", "country_code": "US", "type": "education", "lineage": ["https://openalex.org/I95457486"]}], "countries": ["US"], "is_corresponding": false, "raw_author_name": "Li, Dacheng", "raw_affiliation_strings": ["Department of Electrical Engineering and Computer Sciences , University of California , Berkeley"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Department of Electrical Engineering and Computer Sciences , University of California , Berkeley", "institution_ids": ["https://openalex.org/I95457486"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5039788110", "display_name": "Shiyi Cao", "orcid": "https://orcid.org/0000-0002-1953-7751", "observed_orcids": ["https://orcid.org/0000-0002-1953-7751", "https://orcid.org/0009-0005-6834-375X"]}, "institutions": [{"id": "https://openalex.org/I95457486", "display_name": "University of California, Berkeley", "ror": "https://ror.org/01an7q238", "country_code": "US", "type": "education", "lineage": ["https://openalex.org/I95457486"]}], "countries": ["US"], "is_corresponding": false, "raw_author_name": "Cao, Shiyi", "raw_affiliation_strings": ["Department of Electrical Engineering and Computer Sciences , University of California , Berkeley"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Department of Electrical Engineering and Computer Sciences , University of California , Berkeley", "institution_ids": ["https://openalex.org/I95457486"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5095886747", "display_name": "Tyler Griggs", "orcid": "https://orcid.org/0009-0006-4655-0229", "observed_orcids": ["https://orcid.org/0009-0006-4655-0229"]}, "institutions": [{"id": "https://openalex.org/I95457486", "display_name": "University of California, Berkeley", "ror": "https://ror.org/01an7q238", "country_code": "US", "type": "education", "lineage": ["https://openalex.org/I95457486"]}], "countries": ["US"], "is_corresponding": false, "raw_author_name": "Griggs, Tyler", "raw_affiliation_strings": ["Department of Electrical Engineering and Computer Sciences , University of California , Berkeley"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Department of Electrical Engineering and Computer Sciences , University of California , Berkeley", "institution_ids": ["https://openalex.org/I95457486"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5023175954", "display_name": "Shu Liu", "orcid": "https://orcid.org/0000-0002-9812-7821", "observed_orcids": ["https://orcid.org/0000-0002-9812-7821"]}, "institutions": [{"id": "https://openalex.org/I95457486", "display_name": "University of California, Berkeley", "ror": "https://ror.org/01an7q238", "country_code": "US", "type": "education", "lineage": ["https://openalex.org/I95457486"]}], "countries": ["US"], "is_corresponding": false, "raw_author_name": "Liu, Shu", "raw_affiliation_strings": ["Department of Electrical Engineering and Computer Sciences , University of California , Berkeley"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Department of Electrical Engineering and Computer Sciences , University of California , Berkeley", "institution_ids": ["https://openalex.org/I95457486"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5042410262", "display_name": "Xiangxi Mo", "orcid": null, "observed_orcids": []}, "institutions": [{"id": "https://openalex.org/I95457486", "display_name": "University of California, Berkeley", "ror": "https://ror.org/01an7q238", "country_code": "US", "type": "education", "lineage": ["https://openalex.org/I95457486"]}], "countries": ["US"], "is_corresponding": false, "raw_author_name": "Mo, Xiangxi", "raw_affiliation_strings": ["Department of Electrical Engineering and Computer Sciences , University of California , Berkeley"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Department of Electrical Engineering and Computer Sciences , University of California , Berkeley", "institution_ids": ["https://openalex.org/I95457486"]}]}, {"author_position": "middle", "author": {"id": null, "display_name": "Tang, Eric", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Tang, Eric", "raw_affiliation_strings": ["Anyscale"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Anyscale", "institution_ids": []}]}, {"author_position": "middle", "author": {"id": null, "display_name": "Hegde, Sumanth", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Hegde, Sumanth", "raw_affiliation_strings": ["Anyscale"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Anyscale", "institution_ids": []}]}, {"author_position": "middle", "author": {"id": null, "display_name": "Hakhamaneshi, Kourosh", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Hakhamaneshi, Kourosh", "raw_affiliation_strings": ["Anyscale"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Anyscale", "institution_ids": []}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5022259826", "display_name": "Shishir G. Patil", "orcid": "https://orcid.org/0000-0001-6835-388X", "observed_orcids": ["https://orcid.org/0000-0001-6835-388X"]}, "institutions": [{"id": "https://openalex.org/I95457486", "display_name": "University of California, Berkeley", "ror": "https://ror.org/01an7q238", "country_code": "US", "type": "education", "lineage": ["https://openalex.org/I95457486"]}], "countries": ["US"], "is_corresponding": false, "raw_author_name": "Patil, Shishir G.", "raw_affiliation_strings": ["Department of Electrical Engineering and Computer Sciences , University of California , Berkeley"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Department of Electrical Engineering and Computer Sciences , University of California , Berkeley", "institution_ids": ["https://openalex.org/I95457486"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5005554337", "display_name": "Matei Zaharia", "orcid": "https://orcid.org/0000-0002-7547-7204", "observed_orcids": ["https://orcid.org/0000-0002-7547-7204"]}, "institutions": [{"id": "https://openalex.org/I95457486", "display_name": "University of California, Berkeley", "ror": "https://ror.org/01an7q238", "country_code": "US", "type": "education", "lineage": ["https://openalex.org/I95457486"]}], "countries": ["US"], "is_corresponding": false, "raw_author_name": "Zaharia, Matei", "raw_affiliation_strings": ["Department of Electrical Engineering and Computer Sciences , University of California , Berkeley"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Department of Electrical Engineering and Computer Sciences , University of California , Berkeley", "institution_ids": ["https://openalex.org/I95457486"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5072427753", "display_name": "Joseph E. Gonzalez", "orcid": "https://orcid.org/0000-0003-2921-956X", "observed_orcids": ["https://orcid.org/0000-0003-2921-956X"]}, "institutions": [{"id": "https://openalex.org/I95457486", "display_name": "University of California, Berkeley", "ror": "https://ror.org/01an7q238", "country_code": "US", "type": "education", "lineage": ["https://openalex.org/I95457486"]}], "countries": ["US"], "is_corresponding": false, "raw_author_name": "Gonzalez, Joseph E.", "raw_affiliation_strings": ["Department of Electrical Engineering and Computer Sciences , University of California , Berkeley"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Department of Electrical Engineering and Computer Sciences , University of California , Berkeley", "institution_ids": ["https://openalex.org/I95457486"]}]}, {"author_position": "last", "author": {"id": "https://openalex.org/A5041920173", "display_name": "Ion Gabriel Stoica", "orcid": "https://orcid.org/0000-0002-5373-0088", "observed_orcids": ["https://orcid.org/0000-0002-5373-0088"]}, "institutions": [{"id": "https://openalex.org/I95457486", "display_name": "University of California, Berkeley", "ror": "https://ror.org/01an7q238", "country_code": "US", "type": "education", "lineage": ["https://openalex.org/I95457486"]}], "countries": ["US"], "is_corresponding": false, "raw_author_name": "Stoica, Ion", "raw_affiliation_strings": ["Department of Electrical Engineering and Computer Sciences , University of California , Berkeley"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Department of Electrical Engineering and Computer Sciences , University of California , Berkeley", "institution_ids": ["https://openalex.org/I95457486"]}]}], "abstract_inverted_index": {"Large": [0, 41], "reasoning": [1, 6, 50, 126, 142, 166, 205, 220], "models": [2], "(LRMs)": [3], "tackle": [4], "complex": [5], "problems": [7], "by": [8], "following": [9], "long": [10, 64], "chain-of-thoughts": [11], "(Long": [12], "CoT)": [13], "that": [14, 39, 109, 153], "incorporate": [15], "reflection,": [16], "backtracking,": [17], "and": [18, 24, 56, 80, 89, 103, 209], "self-validation.": [19], "However,": [20], "the": [21, 68, 96, 110, 118, 122, 158, 216, 224], "training": [22, 66, 136, 191, 215], "techniques": [23], "data": [25], "requirements": [26], "to": [27, 95, 117, 190, 203], "elicit": [28, 204], "Long": [29, 48, 113, 159, 177], "CoT": [30, 49, 65, 114, 178], "remain": [31], "poorly": [32], "understood.": [33], "In": [34, 149], "this": [35], "work,": [36], "we": [37, 107], "find": [38, 108], "a": [40, 75, 173], "Language": [42], "model": [43, 70, 174], "(LLM)": [44], "can": [45], "effectively": [46], "learn": [47], "through": [51], "data-efficient": [52], "supervised": [53], "fine-tuning": [54], "(SFT)": [55], "parameter-efficient": [57], "low-rank": [58], "adaptation": [59], "(LoRA).": [60], "With": [61], "just": [62], "17k": [63], "samples,": [67], "Qwen2.5-32B-Instruct": [69], "achieves": [71, 184], "significant": [72], "improvements": [73], "on": [74, 86, 92, 137, 147, 176], "wide": [76], "range": [77], "of": [78, 101, 112, 124, 201, 219, 227], "math": [79], "coding": [81], "benchmarks,": [82], "including": [83], "56.7%": [84], "(+40.0%)": [85], "AIME": [87], "2024": [88], "57.0%": [90], "(+8.1%)": [91], "LiveCodeBench,": [93], "competitive": [94], "proprietary": [97], "o1-preview": [98], "model's": [99], "score": [100], "44.6%": [102], "59.1%.": [104], "More": [105], "importantly,": [106], "structure": [111], "is": [115, 223], "critical": [116], "learning": [119], "process,": [120], "whereas": [121], "content": [123], "individual": [125], "steps": [127], "has": [128], "minimal": [129], "impact.": [130], "Perturbations": [131], "affecting": [132], "content,": [133], "such": [134, 161], "as": [135, 162], "incorrect": [138, 181], "samples": [139, 179], "or": [140, 164], "removing": [141], "keywords,": [143], "have": [144], "little": [145], "impact": [146], "performance.": [148], "contrast,": [150], "structural": [151], "modifications": [152], "disrupt": [154], "logical": [155], "consistency": [156], "in": [157, 207], "CoT,": [160], "shuffling": [163], "deleting": [165], "steps,": [167], "significantly": [168], "degrade": [169], "accuracy.": [170], "For": [171], "example,": [172], "trained": [175], "with": [180, 192], "answers": [182], "still": [183], "only": [185], "3.2%": [186], "lower": [187], "accuracy": [188], "compared": [189], "fully": [193], "correct": [194], "samples.": [195], "These": [196], "insights": [197], "deepen": [198], "our": [199, 228], "understanding": [200], "how": [202], "capabilities": [206], "LLMs": [208], "highlight": [210], "key": [211], "considerations": [212], "for": [213], "efficiently": [214], "next": [217], "generation": [218], "models.": [221], "This": [222], "academic": [225], "paper": [226], "previous": [229], "released": [230], "Sky-T1-32B-Preview": [231], "model.": [232], "Codes": [233], "are": [234], "available": [235], "at": [236], "https://github.com/NovaSky-AI/SkyThought.": [237]}, "primary_topic": {"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.7792999744415283, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.7792999744415283, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10181", "display_name": "Natural Language Processing Techniques", "score": 0.0348999984562397, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T11307", "display_name": "Domain Adaptation and Few-Shot Learning", "score": 0.03020000085234642, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "pmh:oai:arXiv.org:2502.07374", "is_oa": true, "landing_page_url": "http://arxiv.org/abs/2502.07374", "pdf_url": "https://arxiv.org/pdf/2502.07374", "source": {"id": "https://openalex.org/S4306400194", "display_name": "arXiv (Cornell University)", "issn_l": "2331-8422", "issn": ["2331-8422"], "is_oa": true, "is_in_doaj": false, "is_core": false, "listed_in": [], "host_organization": "https://openalex.org/I205783295", "host_organization_name": "Cornell University", "host_organization_lineage": ["https://openalex.org/I205783295"], "host_organization_lineage_names": [], "type": "repository"}, "license": "cc-by", "license_id": "https://openalex.org/licenses/cc-by", "version": "submittedVersion", "is_accepted": false, "is_published": false, "raw_source_name": null, "raw_type": "text"}, "cited_by_count": 0, "is_retracted": false}, {"id": "https://openalex.org/W7103987378", "doi": "https://doi.org/10.1162/tacl.a.44", "title": "Safe Pruning LoRA: Robust Distance-Guided Pruning for Safety Alignment in Adaptation of LLMs", "publication_year": 2025, "publication_date": "2025-01-01", "type": "article", "language": "en", "authorships": [{"author_position": "first", "author": {"id": null, "display_name": "Shuang Ao", "orcid": null, "observed_orcids": []}, "institutions": [{"id": "https://openalex.org/I43439940", "display_name": "University of Southampton", "ror": "https://ror.org/01ryk1543", "country_code": "GB", "type": "education", "lineage": ["https://openalex.org/I43439940"]}], "countries": ["GB"], "is_corresponding": false, "raw_author_name": "Shuang Ao", "raw_affiliation_strings": ["School of Electronics and Computer Science, University of Southampton, UK. s.ao@soton.ac.uk"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "School of Electronics and Computer Science, University of Southampton, UK. s.ao@soton.ac.uk", "institution_ids": ["https://openalex.org/I43439940"]}]}, {"author_position": "middle", "author": {"id": null, "display_name": "Yi Dong", "orcid": null, "observed_orcids": []}, "institutions": [{"id": "https://openalex.org/I146655781", "display_name": "University of Liverpool", "ror": "https://ror.org/04xs57h96", "country_code": "GB", "type": "education", "lineage": ["https://openalex.org/I146655781"]}], "countries": ["GB"], "is_corresponding": false, "raw_author_name": "Yi Dong", "raw_affiliation_strings": ["Department of Computer Science, University of Liverpool, UK. yi.dong@liverpool.ac.uk"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Department of Computer Science, University of Liverpool, UK. yi.dong@liverpool.ac.uk", "institution_ids": ["https://openalex.org/I146655781"]}]}, {"author_position": "middle", "author": {"id": null, "display_name": "Jinwei Hu", "orcid": null, "observed_orcids": []}, "institutions": [{"id": "https://openalex.org/I146655781", "display_name": "University of Liverpool", "ror": "https://ror.org/04xs57h96", "country_code": "GB", "type": "education", "lineage": ["https://openalex.org/I146655781"]}], "countries": ["GB"], "is_corresponding": false, "raw_author_name": "Jinwei Hu", "raw_affiliation_strings": ["Department of Computer Science, University of Liverpool, UK. j.hu33@liverpool.ac.uk"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Department of Computer Science, University of Liverpool, UK. j.hu33@liverpool.ac.uk", "institution_ids": ["https://openalex.org/I146655781"]}]}, {"author_position": "last", "author": {"id": null, "display_name": "Sarvapali D. Ramchurn", "orcid": null, "observed_orcids": []}, "institutions": [{"id": "https://openalex.org/I43439940", "display_name": "University of Southampton", "ror": "https://ror.org/01ryk1543", "country_code": "GB", "type": "education", "lineage": ["https://openalex.org/I43439940"]}], "countries": ["GB"], "is_corresponding": false, "raw_author_name": "Sarvapali D. Ramchurn", "raw_affiliation_strings": ["School of Electronics and Computer Science, University of Southampton, UK. sdr1@soton.ac.uk"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "School of Electronics and Computer Science, University of Southampton, UK. sdr1@soton.ac.uk", "institution_ids": ["https://openalex.org/I43439940"]}]}], "abstract_inverted_index": {"Abstract": [0], "Fine-tuning": [1], "Large": [2], "Language": [3], "Models": [4], "(LLMs)": [5], "with": [6, 23, 100], "Low-Rank": [7], "Adaptation": [8], "(LoRA)": [9], "enhances": [10], "adaptability": [11], "while": [12, 71, 132], "reducing": [13, 129], "computational": [14], "costs.": [15], "However,": [16], "fine-tuning": [17], "can": [18], "compromise": [19], "safety": [20, 32, 67, 70, 88, 125, 130], "alignment,": [21, 68], "even": [22], "benign": [24, 103, 109], "data,": [25, 106], "increasing": [26], "susceptibility": [27], "to": [28, 36, 42], "harmful": [29], "outputs.": [30], "Existing": [31], "alignment": [33, 126], "methods": [34], "struggle": [35], "capture": [37], "complex": [38], "parameter": [39], "shifts,": [40], "leading": [41], "suboptimal": [43], "safety-utility": [44], "trade-offs.": [45], "To": [46], "address": [47], "this": [48], "issue,": [49], "we": [50, 77], "propose": [51], "Safe": [52], "Pruning": [53], "LoRA": [54, 63], "(SPLoRA),": [55], "a": [56, 81, 147], "novel": [57], "pruning-based": [58], "approach": [59], "that": [60, 65, 85, 121], "selectively": [61], "removes": [62], "layers": [64], "weaken": [66], "improving": [69, 135], "preserving": [72], "performance.": [73], "At": [74], "its": [75], "core,": [76], "introduce": [78], "Empirical-DIEM": [79], "(E-DIEM),": [80], "dimension-insensitive": [82], "similarity": [83], "metric": [84], "effectively": [86], "detects": [87], "misalignment": [89], "in": [90], "LoRA-adapted": [91], "models.": [92], "We": [93], "conduct": [94], "extensive": [95], "experiments": [96], "on": [97], "LLMs": [98], "fine-tuned": [99], "mixed": [101], "of": [102], "and": [104, 107, 116, 138, 149, 155], "malicious": [105], "purely": [108], "datasets,": [110], "evaluating": [111], "SPLoRA": [112, 122, 141], "across": [113], "utility,": [114], "safety,": [115], "reliability": [117], "metrics.": [118], "Results": [119], "demonstrate": [120], "outperforms": [123], "state-of-the-art": [124], "techniques,": [127], "significantly": [128], "risks": [131], "maintaining": [133], "or": [134], "model": [136], "performance": [137], "reliability.": [139], "Additionally,": [140], "reduces": [142], "inference": [143], "overhead,": [144], "making": [145], "it": [146], "scalable": [148], "efficient": [150], "solution": [151], "for": [152], "deploying": [153], "safer": [154], "more": [156], "reliable": [157], "LLMs.": [158], "The": [159], "code": [160], "is": [161], "available": [162], "at": [163], "https://github.com/AoShuang92/SPLoRA.": [164]}, "primary_topic": {"id": "https://openalex.org/T11689", "display_name": "Adversarial Robustness in Machine Learning", "score": 0.9204000234603882, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T11689", "display_name": "Adversarial Robustness in Machine Learning", "score": 0.9204000234603882, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.0215000007301569, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10883", "display_name": "Ethics and Social Impacts of AI", "score": 0.013100000098347664, "subfield": {"id": "https://openalex.org/subfields/3311", "display_name": "Safety Research"}, "field": {"id": "https://openalex.org/fields/33", "display_name": "Social Sciences"}, "domain": {"id": "https://openalex.org/domains/2", "display_name": "Social Sciences"}}], "primary_location": {"id": "doi:10.1162/tacl.a.44", "is_oa": true, "landing_page_url": "https://doi.org/10.1162/tacl.a.44", "pdf_url": null, "source": {"id": "https://openalex.org/S2729999759", "display_name": "Transactions of the Association for Computational Linguistics", "issn_l": "2307-387X", "issn": ["2307-387X"], "is_oa": true, "is_in_doaj": true, "is_core": true, "listed_in": ["cwts-core", "doaj", "jufo-1", "norway-2"], "host_organization": "https://openalex.org/P4310320244", "host_organization_name": "Association for Computational Linguistics", "host_organization_lineage": ["https://openalex.org/P4310320244"], "host_organization_lineage_names": ["Association for Computational Linguistics"], "type": "journal"}, "license": "cc-by", "license_id": "https://openalex.org/licenses/cc-by", "version": "publishedVersion", "is_accepted": true, "is_published": true, "raw_source_name": "Transactions of the Association for Computational Linguistics", "raw_type": "journal-article"}, "cited_by_count": 4, "is_retracted": false}, {"id": "https://openalex.org/W7216254808", "doi": "https://doi.org/10.31605/jcis.v8i2.5343", "title": "Sistem Tanya Jawab Layanan Adminduk dengan Retrieval Augmented Generation Komodo-7B", "publication_year": 2025, "publication_date": "2025-11-21", "type": "article", "language": "en", "authorships": [{"author_position": "first", "author": {"id": "https://openalex.org/A5154739268", "display_name": "Adelia Azizatul Haq", "orcid": null, "observed_orcids": []}, "institutions": [{"id": "https://openalex.org/I3130997290", "display_name": "University of Pembangunan Nasional Veteran Jawa Timur", "ror": "https://ror.org/05sbm1c04", "country_code": "ID", "type": "education", "lineage": ["https://openalex.org/I3130997290"]}], "countries": ["ID"], "is_corresponding": false, "raw_author_name": "Haq, Adelia Azizatul", "raw_affiliation_strings": ["a:1:{s:5:\"en_US\";s:22:\"UPN Veteran Jawa Timur\";}"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "a:1:{s:5:\"en_US\";s:22:\"UPN Veteran Jawa Timur\";}", "institution_ids": ["https://openalex.org/I3130997290"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5155687181", "display_name": "I Gede Susrama Mas Diyasa", "orcid": null, "observed_orcids": []}, "institutions": [{"id": "https://openalex.org/I3130997290", "display_name": "University of Pembangunan Nasional Veteran Jawa Timur", "ror": "https://ror.org/05sbm1c04", "country_code": "ID", "type": "education", "lineage": ["https://openalex.org/I3130997290"]}], "countries": ["ID"], "is_corresponding": false, "raw_author_name": "I Gede Susrama Mas Diyasa", "raw_affiliation_strings": ["UPN Veteran Jawa Timur"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "UPN Veteran Jawa Timur", "institution_ids": ["https://openalex.org/I3130997290"]}]}, {"author_position": "last", "author": {"id": "https://openalex.org/A5156219211", "display_name": "Sugiarto", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Sugiarto", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}], "abstract_inverted_index": {"Pesatnya": [0], "transformasi": [1], "digital": [2], "di": [3, 16, 47], "Indonesia": [4], "mendorong": [5], "pemerintah": [6], "untuk": [7, 68], "terus": [8], "berinovasi": [9], "dalam": [10, 36], "meningkatkan": [11, 53, 105], "kualitas": [12, 54], "layanan": [13, 32, 41, 72], "publik,": [14], "termasuk": [15], "bidang": [17], "administrasi": [18], "kependudukan.": [19], "Dinas": [20], "Kependudukan": [21], "dan": [22, 45, 77, 98, 107, 118, 131, 143, 157], "Pencatatan": [23], "Sipil": [24], "(Disdukcapil)": [25], "Kota": [26, 116], "Surabaya": [27, 117], "sebagai": [28], "salah": [29], "satu": [30], "penyedia": [31], "kependudukan,": [33], "menghadapi": [34], "tantangan": [35], "menjawab": [37, 69], "kebutuhan": [38], "masyarakat": [39], "akan": [40], "yang": [42, 88], "cepat,": [43], "responsif,": [44], "tersedia": [46], "luar": [48], "jam": [49], "operasional.": [50], "Dalam": [51], "rangka": [52], "layanan,": [55], "penelitian": [56], "ini": [57], "mengembangkan": [58], "sistem": [59], "Question": [60], "Answering": [61], "(QA)": [62], "berbasis": [63, 134], "Large": [64], "Language": [65], "Model": [66], "(LLM)": [67], "pertanyaan": [70], "seputar": [71], "Kartu": [73, 78], "Tanda": [74], "Penduduk": [75], "(KTP)": [76], "Keluarga": [79], "(KK).": [80], "Sistem": [81], "dirancang": [82], "dengan": [83, 128], "memanfaatkan": [84], "model": [85], "LLM": [86], "Komodo-7B": [87, 149], "telah": [89], "disesuaikan": [90], "menggunakan": [91, 125, 139], "teknik": [92], "fine-tuning": [93, 148], "Quantized": [94], "Low-Rank": [95], "Adaptation": [96], "(QloRA)": [97], "pendekatan": [99], "Retrieval": [100], "Augmented": [101], "Generation": [102], "(RAG)": [103], "guna": [104], "akurasi": [106], "relevansi": [108], "jawaban.": [109], "Data": [110], "pelatihan": [111], "mencakup": [112], "dataset": [113], "pengaduan-pertanyaan": [114], "Disdukcapil": [115], "empat": [119], "data": [120], "open": [121], "source.": [122], "Proses": [123], "RAG": [124, 147], "vektorisasi": [126], "kalimat": [127], "sentence": [129], "transformer": [130], "pemanggilan": [132], "konteks": [133], "cosine": [135], "similarity.": [136], "Evaluasi": [137], "dilakukan": [138], "metrik": [140], "ROUGE-1,": [141], "ROUGE-L,": [142], "METEOR.": [144], "Hasil": [145], "evaluasi": [146], "didapatkan": [150], "F1-Score": [151], "ROUGE-1": [152], "mencapai": [153], "0.1243,": [154], "ROUGE-L": [155], "0.0854,": [156], "skor": [158], "METEOR": [159], "0.2013.": [160]}, "primary_topic": {"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.9848999977111816, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.9848999977111816, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10181", "display_name": "Natural Language Processing Techniques", "score": 0.008299999870359898, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T12031", "display_name": "Speech and dialogue systems", "score": 0.001500000013038516, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "pmh:oai:ojs.unsulbar.ac.id:article/5343", "is_oa": true, "landing_page_url": "https://ojs.unsulbar.ac.id/index.php/jcis/article/view/5343", "pdf_url": null, "source": {"id": "https://openalex.org/S4210227097", "display_name": "Journal of Computer and Information System ( J-CIS )", "issn_l": "2622-0881", "issn": ["2622-0881", "2622-5859"], "is_oa": true, "is_in_doaj": false, "is_core": false, "listed_in": [], "host_organization": "https://openalex.org/P4404673986", "host_organization_name": "West Sulawesi University", "host_organization_lineage": ["https://openalex.org/P4404673986"], "host_organization_lineage_names": ["West Sulawesi University"], "type": "journal"}, "license": null, "license_id": null, "version": "publishedVersion", "is_accepted": true, "is_published": true, "raw_source_name": "10.31605/jcis.v8i2", "raw_type": "info:eu-repo/semantics/article"}, "cited_by_count": 0, "is_retracted": false}, {"id": "https://openalex.org/W4407761215", "doi": "https://doi.org/10.3389/fpubh.2025.1540946", "title": "Building an intelligent diabetes Q&A system with knowledge graphs and large language models", "publication_year": 2025, "publication_date": "2025-02-20", "type": "article", "language": "en", "authorships": [{"author_position": "first", "author": {"id": null, "display_name": "Zhenkai Qin", "orcid": null, "observed_orcids": []}, "institutions": [{"id": "https://openalex.org/I4405280785", "display_name": "Guangxi Police College", "ror": "https://ror.org/03ef1fb09", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I4405280785"]}, {"id": "https://openalex.org/I4800084", "display_name": "Southwest Jiaotong University", "ror": "https://ror.org/00hn7w693", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I4800084"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Zhenkai Qin", "raw_affiliation_strings": ["School of Computer Science and Artificial Intelligence, Southwest Jiaotong University, Chengdu, China", "School of Information Technology, Guangxi Police College, Nanning, China"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "School of Computer Science and Artificial Intelligence, Southwest Jiaotong University, Chengdu, China", "institution_ids": ["https://openalex.org/I4800084"]}, {"raw_affiliation_string": "School of Information Technology, Guangxi Police College, Nanning, China", "institution_ids": ["https://openalex.org/I4405280785"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5100771814", "display_name": "Dongze Wu", "orcid": "https://orcid.org/0000-0002-5571-8728", "observed_orcids": ["https://orcid.org/0000-0002-5571-8728", "https://orcid.org/0009-0000-8053-1335"]}, "institutions": [{"id": "https://openalex.org/I4405280785", "display_name": "Guangxi Police College", "ror": "https://ror.org/03ef1fb09", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I4405280785"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Dongze Wu", "raw_affiliation_strings": ["School of Information Technology, Guangxi Police College, Nanning, China"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "School of Information Technology, Guangxi Police College, Nanning, China", "institution_ids": ["https://openalex.org/I4405280785"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5108241136", "display_name": "Zhidong Zang", "orcid": "https://orcid.org/0000-0002-4165-7178", "observed_orcids": ["https://orcid.org/0000-0002-4165-7178"]}, "institutions": [{"id": "https://openalex.org/I78978612", "display_name": "Yangzhou University", "ror": "https://ror.org/03tqb8s11", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I78978612"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Zhidong Zang", "raw_affiliation_strings": ["School of Social Development, Yangzhou University, Yangzhou, China"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "School of Social Development, Yangzhou University, Yangzhou, China", "institution_ids": ["https://openalex.org/I78978612"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5100365891", "display_name": "Xiaolong Chen", "orcid": "https://orcid.org/0000-0002-8704-8212", "observed_orcids": ["https://orcid.org/0000-0002-8704-8212"]}, "institutions": [{"id": "https://openalex.org/I49835588", "display_name": "Macao Polytechnic University", "ror": "https://ror.org/02sf5td35", "country_code": "MO", "type": "education", "lineage": ["https://openalex.org/I49835588"]}], "countries": ["MO"], "is_corresponding": true, "raw_author_name": "Xiaolong Chen", "raw_affiliation_strings": ["Faculty of Humanities and Social Sciences, Macao Polytechnic University, Macao, China"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Faculty of Humanities and Social Sciences, Macao Polytechnic University, Macao, China", "institution_ids": ["https://openalex.org/I49835588"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5100714194", "display_name": "Hongfeng Zhang", "orcid": "https://orcid.org/0000-0002-7586-2364", "observed_orcids": ["https://orcid.org/0000-0002-7586-2364"]}, "institutions": [{"id": "https://openalex.org/I49835588", "display_name": "Macao Polytechnic University", "ror": "https://ror.org/02sf5td35", "country_code": "MO", "type": "education", "lineage": ["https://openalex.org/I49835588"]}], "countries": ["MO"], "is_corresponding": true, "raw_author_name": "Hongfeng Zhang", "raw_affiliation_strings": ["Faculty of Humanities and Social Sciences, Macao Polytechnic University, Macao, China"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Faculty of Humanities and Social Sciences, Macao Polytechnic University, Macao, China", "institution_ids": ["https://openalex.org/I49835588"]}]}, {"author_position": "last", "author": {"id": "https://openalex.org/A5080595813", "display_name": "Cora Un In Wong", "orcid": "https://orcid.org/0000-0001-5523-4355", "observed_orcids": ["https://orcid.org/0000-0001-5523-4355"]}, "institutions": [{"id": "https://openalex.org/I49835588", "display_name": "Macao Polytechnic University", "ror": "https://ror.org/02sf5td35", "country_code": "MO", "type": "education", "lineage": ["https://openalex.org/I49835588"]}], "countries": ["MO"], "is_corresponding": false, "raw_author_name": "Cora Un In Wong", "raw_affiliation_strings": ["Faculty of Humanities and Social Sciences, Macao Polytechnic University, Macao, China"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Faculty of Humanities and Social Sciences, Macao Polytechnic University, Macao, China", "institution_ids": ["https://openalex.org/I49835588"]}]}], "abstract_inverted_index": {"Introduction:": [0], "This": [1, 142], "paper": [2], "introduces": [3], "an": [4], "intelligent": [5], "question-answering": [6, 159], "system": [7, 26, 51, 102], "designed": [8], "to": [9, 14, 28, 84, 133, 156], "deliver": [10], "personalized": [11, 135, 182], "medical": [12, 35, 47, 136, 158], "information": [13], "diabetic": [15], "patients.": [16], "By": [17], "integrating": [18, 148], "large": [19, 149], "language": [20, 150], "models": [21, 151], "with": [22, 57, 152], "knowledge": [23, 55, 120, 154], "graphs,": [24], "the": [25, 38, 58, 78, 124, 145], "aims": [27], "provide": [29, 134], "more": [30], "accurate": [31], "and": [32, 60, 69, 82, 96, 109, 127, 172], "contextually": [33], "relevant": [34], "guidance,": [36], "addressing": [37], "limitations": [39], "of": [40, 117, 147], "traditional": [41], "healthcare": [42, 174, 183], "systems": [43], "in": [44, 106, 112], "handling": [45], "complex": [46], "queries.": [48], "Methods:": [49], "The": [50, 88, 101, 115, 161], "combines": [52], "a": [53, 118, 165, 177], "Neo4j-based": [54], "graph": [56, 121], "Baichuan2-13B": [59], "Qwen2.5-7B": [61], "models.": [62], "To": [63], "enhance": [64], "performance,": [65], "Low-Rank": [66], "Adaptation": [67], "(LoRA)": [68], "prompt-based": [70], "learning": [71], "techniques": [72], "are": [73], "applied.": [74], "These": [75], "methods": [76], "improve": [77, 157], "system's": [79, 89, 125], "semantic": [80], "understanding": [81], "ability": [83, 132], "generate": [85], "high-quality": [86], "responses.": [87], "performance": [90], "is": [91], "evaluated": [92], "using": [93], "entity": [94, 107], "recognition": [95, 108], "intent": [97, 113], "classification": [98], "tasks.": [99], "Results:": [100], "achieves": [103], "85.91%": [104], "precision": [105, 111], "88.55%": [110], "classification.": [114], "integration": [116], "structured": [119, 153], "significantly": [122], "improves": [123], "accuracy": [126], "clinical": [128], "relevance,": [129], "enhancing": [130], "its": [131], "responses": [137], "for": [138, 168, 180], "diabetes": [139, 170], "management.": [140], "Discussion:": [141], "study": [143], "demonstrates": [144], "effectiveness": [146], "graphs": [155], "systems.": [160], "proposed": [162], "approach": [163], "offers": [164], "promising": [166], "framework": [167], "advancing": [169], "management": [171], "other": [173], "applications,": [175], "providing": [176], "solid": [178], "foundation": [179], "future": [181], "interventions.": [184]}, "primary_topic": {"id": "https://openalex.org/T11636", "display_name": "Artificial Intelligence in Healthcare and Education", "score": 0.8852999806404114, "subfield": {"id": "https://openalex.org/subfields/2718", "display_name": "Health Informatics"}, "field": {"id": "https://openalex.org/fields/27", "display_name": "Medicine"}, "domain": {"id": "https://openalex.org/domains/4", "display_name": "Health Sciences"}}, "topics": [{"id": "https://openalex.org/T11636", "display_name": "Artificial Intelligence in Healthcare and Education", "score": 0.8852999806404114, "subfield": {"id": "https://openalex.org/subfields/2718", "display_name": "Health Informatics"}, "field": {"id": "https://openalex.org/fields/27", "display_name": "Medicine"}, "domain": {"id": "https://openalex.org/domains/4", "display_name": "Health Sciences"}}, {"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.025299999862909317, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T11710", "display_name": "Biomedical Text Mining and Ontologies", "score": 0.020400000736117363, "subfield": {"id": "https://openalex.org/subfields/1312", "display_name": "Molecular Biology"}, "field": {"id": "https://openalex.org/fields/13", "display_name": "Biochemistry, Genetics and Molecular Biology"}, "domain": {"id": "https://openalex.org/domains/1", "display_name": "Life Sciences"}}], "primary_location": {"id": "doi:10.3389/fpubh.2025.1540946", "is_oa": true, "landing_page_url": "https://doi.org/10.3389/fpubh.2025.1540946", "pdf_url": "https://www.frontiersin.org/journals/public-health/articles/10.3389/fpubh.2025.1540946/pdf", "source": {"id": "https://openalex.org/S2595931848", "display_name": "Frontiers in Public Health", "issn_l": "2296-2565", "issn": ["2296-2565"], "is_oa": true, "is_in_doaj": true, "is_core": true, "listed_in": ["cwts-core", "doaj", "medline", "norway-1"], "host_organization": "https://openalex.org/P4310320527", "host_organization_name": "Frontiers Media", "host_organization_lineage": ["https://openalex.org/P4310320527"], "host_organization_lineage_names": ["Frontiers Media"], "type": "journal"}, "license": "cc-by", "license_id": "https://openalex.org/licenses/cc-by", "version": "publishedVersion", "is_accepted": true, "is_published": true, "raw_source_name": "Frontiers in Public Health", "raw_type": "journal-article"}, "cited_by_count": 9, "is_retracted": false}, {"id": "https://openalex.org/W4417286657", "doi": "https://doi.org/10.48550/arxiv.2503.11617", "title": "ASMA-Tune: Unlocking LLMs' Assembly Code Comprehension via Structural-Semantic Instruction Tuning", "publication_year": 2025, "publication_date": "2025-03-14", "type": "preprint", "language": "en", "authorships": [{"author_position": "first", "author": {"id": null, "display_name": "Wang, Xinyi", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Wang, Xinyi", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5059024183", "display_name": "Jiashui Wang", "orcid": "https://orcid.org/0009-0005-3100-0534", "observed_orcids": ["https://orcid.org/0009-0005-3100-0534"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Wang, Jiashui", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": null, "display_name": "Su, Jinbo", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Su, Jinbo", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5100360160", "display_name": "Ke Wang", "orcid": "https://orcid.org/0000-0003-4508-1302", "observed_orcids": ["https://orcid.org/0000-0003-4508-1302", "https://orcid.org/0000-0001-8675-7220", "https://orcid.org/0000-0002-5372-5596", "https://orcid.org/0009-0000-1076-5814"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Wang, Ke", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5102027852", "display_name": "Peng Chen", "orcid": "https://orcid.org/0009-0006-4802-2382", "observed_orcids": ["https://orcid.org/0009-0006-4802-2382"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Chen, Peng", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5107947108", "display_name": "Yanming Liu", "orcid": "https://orcid.org/0000-0002-5522-1682", "observed_orcids": ["https://orcid.org/0000-0002-5522-1682"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Liu, Yanming", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5057532144", "display_name": "Long Liu", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Liu, Long", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5100674281", "display_name": "Xiang Li", "orcid": "https://orcid.org/0000-0002-6482-2535", "observed_orcids": ["https://orcid.org/0000-0002-6482-2535", "https://orcid.org/0000-0002-0434-3057", "https://orcid.org/0000-0002-4738-0034"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Li, Xiang", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": null, "display_name": "Wang, Yangdong", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Wang, Yangdong", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5101830428", "display_name": "Qiyuan Chen", "orcid": "https://orcid.org/0009-0009-0701-8784", "observed_orcids": ["https://orcid.org/0009-0009-0701-8784"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Chen, Qiyuan", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5052511548", "display_name": "R. H. Chen", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Chen, Rongze", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "last", "author": {"id": "https://openalex.org/A5005646333", "display_name": "Chunfu Jia", "orcid": "https://orcid.org/0000-0002-5588-9690", "observed_orcids": ["https://orcid.org/0000-0002-5588-9690"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Jia, Chunfu", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}], "abstract_inverted_index": {"Assembly": [0, 65], "code": [1, 156], "analysis": [2, 56], "and": [3, 23, 101, 124, 138, 144, 147], "comprehension": [4, 120], "play": [5], "critical": [6], "roles": [7], "in": [8, 54, 118], "applications": [9], "like": [10], "reverse": [11], "engineering,": [12], "yet": [13, 57], "they": [14], "face": [15], "substantial": [16], "challenges": [17], "due": [18], "to": [19, 160], "low": [20], "information": [21], "density": [22], "a": [24, 82], "lack": [25], "of": [26], "explicit": [27], "syntactic": [28], "structures.": [29], "While": [30], "traditional": [31], "masked": [32], "language": [33, 43, 48, 107], "modeling": [34], "(MLM)": [35], "approaches": [36], "do": [37], "not": [38], "explicitly": [39], "focus": [40], "on": [41, 141], "natural": [42, 106], "interaction,": [44], "emerging": [45], "decoder-focused": [46], "large": [47], "models": [49, 135], "(LLMs)": [50], "demonstrate": [51, 111], "partial": [52], "success": [53], "binary": [55], "remain": [58], "underexplored": [59], "for": [60], "holistic": [61], "comprehension.": [62], "We": [63], "present": [64], "Augmented": [66], "Tuning,": [67], "an": [68], "end-to-end": [69], "structural-semantic": [70], "instruction": [71], "tuning": [72], "framework": [73], "that": [74], "synergizes": [75], "encoder": [76, 88], "architecture": [77], "with": [78, 97, 121, 154], "decoder-based": [79], "LLMs": [80], "through": [81], "projector": [83, 94], "module,": [84], "where": [85], "the": [86, 93, 98, 102], "assembly": [87, 119], "extracts": [89], "hardware-level": [90], "structural": [91], "features,": [92], "bridges": [95], "representations": [96], "semantic": [99], "space,": [100], "instruction-tuned": [103], "LLM": [104], "preserves": [105], "capabilities.": [108], "Experimental": [109], "results": [110], "three": [112], "key": [113], "advantages:": [114], "(1)": [115], "State-of-the-art": [116], "performance": [117], "+39.7%": [122], "Recall@1": [123, 137], "+17.8%": [125], "MRR": [126, 140], "improvements": [127], "over": [128], "GPT-4-Turbo,": [129], "(2)": [130], "Consistent": [131], "enhancements": [132], "across": [133, 162], "base": [134], "(24.6-107.4%": [136], "15.2-106.3%": [139], "Qwen2.5-Coder,": [142], "Deepseek-Coder": [143], "CodeLlama": [145], "variants),": [146], "(3)": [148], "Superior": [149], "instruction-following": [150], "capabilities": [151], "(41.5%-118%": [152], "improvements)": [153], "controlled": [155], "generation": [157], "degradation": [158], "(-8.9%": [159], "-35%": [161], "architectures).": [163]}, "primary_topic": {"id": "https://openalex.org/T10260", "display_name": "Software Engineering Research", "score": 0.9887999892234802, "subfield": {"id": "https://openalex.org/subfields/1710", "display_name": "Information Systems"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T10260", "display_name": "Software Engineering Research", "score": 0.9887999892234802, "subfield": {"id": "https://openalex.org/subfields/1710", "display_name": "Information Systems"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T11424", "display_name": "Security and Verification in Computing", "score": 0.004100000020116568, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.00279999990016222, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "pmh:oai:arXiv.org:2503.11617", "is_oa": true, "landing_page_url": "http://arxiv.org/abs/2503.11617", "pdf_url": "https://arxiv.org/pdf/2503.11617", "source": {"id": "https://openalex.org/S4306400194", "display_name": "arXiv (Cornell University)", "issn_l": "2331-8422", "issn": ["2331-8422"], "is_oa": true, "is_in_doaj": false, "is_core": false, "listed_in": [], "host_organization": "https://openalex.org/I205783295", "host_organization_name": "Cornell University", "host_organization_lineage": ["https://openalex.org/I205783295"], "host_organization_lineage_names": [], "type": "repository"}, "license": null, "license_id": null, "version": "submittedVersion", "is_accepted": false, "is_published": false, "raw_source_name": null, "raw_type": "text"}, "cited_by_count": 0, "is_retracted": false}, {"id": "https://openalex.org/W4412947727", "doi": "https://doi.org/10.26599/tst.2025.9010106", "title": "FT-HashRAG: Combining Hash Retrieval-Augmented Generation with Fine-Tuning for Universal Explainable Recommendation", "publication_year": 2025, "publication_date": "2025-08-05", "type": "article", "language": "en", "authorships": [{"author_position": "first", "author": {"id": "https://openalex.org/A5100321165", "display_name": "Zixuan Zhang", "orcid": "https://orcid.org/0000-0003-1679-5322", "observed_orcids": ["https://orcid.org/0000-0003-1679-5322"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Zixuan Zhang", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5086511524", "display_name": "Bowen Hao", "orcid": "https://orcid.org/0009-0005-3523-4346", "observed_orcids": ["https://orcid.org/0009-0005-3523-4346", "https://orcid.org/0000-0002-0719-646X"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Bowen Hao", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": null, "display_name": "Yurui Wang", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Yurui Wang", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5062354305", "display_name": "Wei Xiang", "orcid": "https://orcid.org/0000-0002-8756-2211", "observed_orcids": ["https://orcid.org/0000-0002-8756-2211"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Xiang Wei", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": null, "display_name": "Qimeng Niu", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Qimeng Niu", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "last", "author": {"id": "https://openalex.org/A5078396805", "display_name": "Yuxuan Li", "orcid": "https://orcid.org/0000-0002-2670-2006", "observed_orcids": ["https://orcid.org/0000-0002-2670-2006"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Yuxuan Li", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}], "abstract_inverted_index": {"Abstract": [0], "Explainable": [1], "recommendation": [2, 31, 79, 132], "have": [3], "garnered": [4], "significant": [5], "research": [6], "interest": [7], "for": [8, 48], "their": [9], "transparency.": [10], "Recent": [11], "advancements": [12], "employ": [13, 41], "Large": [14], "Language": [15], "Models": [16], "(LLMs)": [17], "through": [18], "two": [19], "primary": [20], "approaches:": [21], "LLM-auxiliary": [22], "models,": [23, 39], "which": [24, 40], "improve": [25], "accuracy": [26], "but": [27], "rely": [28], "on": [29, 166], "conventional": [30], "models": [32], "and": [33, 37, 55, 98, 140, 143, 150], "produce": [34], "limited": [35], "explanations,": [36], "LLM-based": [38], "fine-tuning": [42], "or": [43], "Graph": [44], "Retrieval-Augmented": [45, 84], "Generation": [46, 85], "(GraphRAG)": [47], "explainability\u2014\u2014yet": [49], "face": [50], "limitations": [51], "in": [52], "domain": [53], "adaptability": [54], "neglect": [56], "intrinsic": [57, 94], "signals": [58, 100], "like": [59], "item": [60, 96, 102, 134, 136], "weights.": [61], "Moreover,": [62], "existing": [63], "methods": [64], "struggle": [65], "with": [66, 87, 111], "multi-scenario": [67, 126], "generalization.": [68], "To": [69], "address": [70], "these": [71], "challenges,": [72], "we": [73], "propose": [74], "FT-HashRAG,": [75], "a": [76, 125], "universal": [77], "explainable": [78, 131], "framework": [80], "that": [81], "integrates": [82], "Hash": [83], "(HashRAG)": [86], "fine-tuning.": [88], "HashRAG": [89, 110], "constructs": [90], "user-specific": [91], "blocks": [92], "using": [93], "(e.g.,": [95, 101], "weights)": [97], "extrinsic": [99], "descriptions)": [103], "to": [104, 118], "generate": [105], "fine-grained": [106], "explanations.": [107], "By": [108], "integrating": [109], "fine-tuning,": [112], "FT-HashRAG": [113, 157], "enhances": [114], "the": [115], "LLM\u2019s": [116], "capacity": [117], "synthesize": [119], "retrieved": [120], "information": [121], "effectively.": [122], "We": [123, 162], "curate": [124], "instruction": [127], "dataset": [128], "including": [129], "four": [130], "scenarios:": [133], "weights,": [135], "attributes,": [137], "popularity": [138], "bias,": [139], "interaction": [141], "paths,": [142], "train": [144], "LLM": [145], "via": [146], "Supervised": [147], "Fine-Tuning": [148], "(SFT)": [149], "Direct": [151], "Preference": [152], "Optimization": [153], "(DPO).": [154], "Experiments": [155], "show": [156], "significantly": [158], "outperforms": [159], "state-of-the-art": [160], "baselines.": [161], "open-source": [163], "our": [164], "model": [165], "https://huggingface.co/tczzx6/FT-HashRAG.": [167]}, "primary_topic": {"id": "https://openalex.org/T10203", "display_name": "Recommender Systems and Techniques", "score": 0.9970999956130981, "subfield": {"id": "https://openalex.org/subfields/1710", "display_name": "Information Systems"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T10203", "display_name": "Recommender Systems and Techniques", "score": 0.9970999956130981, "subfield": {"id": "https://openalex.org/subfields/1710", "display_name": "Information Systems"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T12026", "display_name": "Explainable Artificial Intelligence (XAI)", "score": 0.002300000051036477, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10286", "display_name": "Information Retrieval and Search Behavior", "score": 9.999999747378752e-05, "subfield": {"id": "https://openalex.org/subfields/1710", "display_name": "Information Systems"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "doi:10.26599/tst.2025.9010106", "is_oa": true, "landing_page_url": "https://doi.org/10.26599/tst.2025.9010106", "pdf_url": null, "source": {"id": "https://openalex.org/S24978797", "display_name": "Tsinghua Science & Technology", "issn_l": "1007-0214", "issn": ["1007-0214", "1878-7606"], "is_oa": true, "is_in_doaj": true, "is_core": true, "listed_in": ["cwts-core", "doaj"], "host_organization": "https://openalex.org/P4310311901", "host_organization_name": "Tsinghua University Press", "host_organization_lineage": ["https://openalex.org/P4310311901", "https://openalex.org/P4353108601"], "host_organization_lineage_names": ["Tsinghua University Press", "Tsinghua University"], "type": "journal"}, "license": "cc-by", "license_id": "https://openalex.org/licenses/cc-by", "version": "publishedVersion", "is_accepted": true, "is_published": true, "raw_source_name": "Tsinghua Science and Technology", "raw_type": "journal-article"}, "cited_by_count": 0, "is_retracted": false}, {"id": "https://openalex.org/W4415292013", "doi": "https://doi.org/10.36676/urr.v12.i4.1621", "title": "Low-Resource Fine-Tuning of LLMs for Domain-Specific Tasks", "publication_year": 2025, "publication_date": "2025-10-12", "type": "article", "language": "en", "authorships": [{"author_position": "first", "author": {"id": "https://openalex.org/A5120039399", "display_name": "Vamshikrishna Challa", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Vamshikrishna Challa", "raw_affiliation_strings": ["Master's Computer Science Department: Computational Analysis and Modeling (CAM) Email: vamshikrishnachalla1999@gmail.com,  Ruston, Louisiana,  USA"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Master's Computer Science Department: Computational Analysis and Modeling (CAM) Email: vamshikrishnachalla1999@gmail.com,  Ruston, Louisiana,  USA", "institution_ids": []}]}, {"author_position": "last", "author": {"id": null, "display_name": "Abika Osinachi Bright", "orcid": null, "observed_orcids": []}, "institutions": [{"id": "https://openalex.org/I919208787", "display_name": "Louisiana Tech University", "ror": "https://ror.org/04q9esz89", "country_code": "US", "type": "education", "lineage": ["https://openalex.org/I2799628689", "https://openalex.org/I919208787"]}], "countries": ["US"], "is_corresponding": false, "raw_author_name": "Abika Osinachi Bright", "raw_affiliation_strings": ["Masters Electrical Engineering  Department: Material and infrastructural systems engineering  Email: abikab2022@gmail.com ,  Ruston, Louisiana ,  USA"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Masters Electrical Engineering  Department: Material and infrastructural systems engineering  Email: abikab2022@gmail.com ,  Ruston, Louisiana ,  USA", "institution_ids": ["https://openalex.org/I919208787"]}]}], "abstract_inverted_index": {"Large": [0], "Language": [1], "Models": [2], "(LLMs)": [3], "such": [4, 38, 92], "as": [5, 93], "GPT": [6], "and": [7, 24, 33, 49, 62, 95, 119, 126, 133, 144], "LLaMA": [8], "have": [9], "demonstrated": [10], "remarkable": [11], "capabilities": [12], "across": [13], "diverse": [14], "natural": [15], "language": [16], "processing": [17], "(NLP)": [18], "applications.": [19], "However,": [20], "their": [21], "enormous": [22], "computational": [23], "memory": [25], "requirements": [26], "hinder": [27], "adoption": [28], "by": [29], "smaller": [30], "research": [31, 103], "labs": [32], "enterprises.": [34], "Full-scale": [35], "fine-tuning": [36, 106], "of": [37, 111, 116], "models": [39], "is": [40, 100], "often": [41], "infeasible": [42], "due": [43], "to": [44, 70, 89], "high": [45], "GPU": [46], "memory,": [47], "storage,": [48], "energy": [50], "consumption.": [51], "Parameter-Efficient": [52], "Fine-Tuning": [53], "(PEFT)": [54], "techniques,": [55], "including": [56], "Low-Rank": [57], "Adaptation": [58], "(LoRA),": [59], "adapter-based": [60, 134], "methods,": [61], "prefix-tuning,": [63], "present": [64], "an": [65], "alternative": [66], "for": [67, 80, 108, 151], "adapting": [68], "LLMs": [69], "downstream": [71], "tasks": [72], "under": [73], "constrained": [74, 117], "budgets.": [75], "Despite": [76], "progress": [77], "in": [78], "PEFT": [79], "general": [81], "NLP": [82], "benchmarks,": [83], "limited": [84], "attention": [85], "has": [86], "been": [87], "given": [88], "domain-specific": [90], "applications": [91], "healthcare": [94], "energy,": [96], "where": [97], "specialized": [98], "knowledge": [99], "critical.": [101], "This": [102], "investigates": [104], "low-resource": [105], "strategies": [107], "domain": [109], "adaptation": [110], "LLMs,": [112], "identifies": [113], "the": [114], "challenges": [115], "environments,": [118], "evaluates": [120], "practical": [121], "frameworks": [122], "that": [123, 131], "balance": [124], "efficiency": [125], "performance.": [127], "Experimental": [128], "results": [129], "demonstrate": [130], "LoRA-": [132], "methods": [135], "achieve": [136], "competitive": [137], "accuracy": [138], "while": [139], "drastically": [140], "reducing": [141], "trainable": [142], "parameters": [143], "compute": [145], "costs,": [146], "making": [147], "them": [148], "highly": [149], "suitable": [150], "resource-limited": [152], "settings.": [153]}, "primary_topic": {"id": "https://openalex.org/T11307", "display_name": "Domain Adaptation and Few-Shot Learning", "score": 0.8715999722480774, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T11307", "display_name": "Domain Adaptation and Few-Shot Learning", "score": 0.8715999722480774, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.11999999731779099, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10181", "display_name": "Natural Language Processing Techniques", "score": 0.004800000227987766, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "doi:10.36676/urr.v12.i4.1621", "is_oa": true, "landing_page_url": "https://doi.org/10.36676/urr.v12.i4.1621", "pdf_url": "https://urr.shodhsagar.com/index.php/j/article/download/1621/1736", "source": {"id": "https://openalex.org/S4387291108", "display_name": "Universal Research Reports", "issn_l": "2348-5612", "issn": ["2348-5612"], "is_oa": true, "is_in_doaj": false, "is_core": false, "listed_in": [], "host_organization": null, "host_organization_name": null, "host_organization_lineage": [], "host_organization_lineage_names": [], "type": "journal"}, "license": "cc-by-nc", "license_id": "https://openalex.org/licenses/cc-by-nc", "version": "publishedVersion", "is_accepted": true, "is_published": true, "raw_source_name": "Universal Research Reports", "raw_type": "journal-article"}, "cited_by_count": 3, "is_retracted": false}, {"id": "https://openalex.org/W4409039368", "doi": "https://doi.org/10.2139/ssrn.5200113", "title": "Optimizing Large Vision-Language Models for Context-Aware Construction Safety Assessment", "publication_year": 2025, "publication_date": "2025-01-01", "type": "preprint", "language": "en", "authorships": [{"author_position": "first", "author": {"id": "https://openalex.org/A5003452999", "display_name": "Taegeon Kim", "orcid": "https://orcid.org/0000-0003-1682-6061", "observed_orcids": ["https://orcid.org/0000-0003-1682-6061", "https://orcid.org/0000-0001-7815-5887", "https://orcid.org/0009-0003-4770-163X"]}, "institutions": [{"id": "https://openalex.org/I193775966", "display_name": "Yonsei University", "ror": "https://ror.org/01wjejq96", "country_code": "KR", "type": "education", "lineage": ["https://openalex.org/I193775966"]}], "countries": ["KR"], "is_corresponding": false, "raw_author_name": "Taegeon Kim", "raw_affiliation_strings": ["Yonsei University, Seoul, Korea, Republic of (South Korea)"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Yonsei University, Seoul, Korea, Republic of (South Korea)", "institution_ids": ["https://openalex.org/I193775966"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5082880112", "display_name": "Seokhwan Kim", "orcid": "https://orcid.org/0000-0002-7443-1212", "observed_orcids": ["https://orcid.org/0000-0002-7443-1212"]}, "institutions": [{"id": "https://openalex.org/I193775966", "display_name": "Yonsei University", "ror": "https://ror.org/01wjejq96", "country_code": "KR", "type": "education", "lineage": ["https://openalex.org/I193775966"]}], "countries": ["KR"], "is_corresponding": false, "raw_author_name": "Seokhwan Kim", "raw_affiliation_strings": ["Yonsei University, Seoul, Korea, Republic of (South Korea)"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Yonsei University, Seoul, Korea, Republic of (South Korea)", "institution_ids": ["https://openalex.org/I193775966"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5037093946", "display_name": "Wei\u2010Chih Chern", "orcid": "https://orcid.org/0009-0004-1535-0094", "observed_orcids": ["https://orcid.org/0009-0004-1535-0094"]}, "institutions": [{"id": "https://openalex.org/I127591826", "display_name": "University of Dayton", "ror": "https://ror.org/021v3qy27", "country_code": "US", "type": "education", "lineage": ["https://openalex.org/I127591826"]}], "countries": ["US"], "is_corresponding": false, "raw_author_name": "Wei-Chih Chern", "raw_affiliation_strings": ["University of Dayton, 300 College Park, Dayton, OH 45469, United States"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "University of Dayton, 300 College Park, Dayton, OH 45469, United States", "institution_ids": ["https://openalex.org/I127591826"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5101950079", "display_name": "Somin Park", "orcid": "https://orcid.org/0000-0002-8885-0872", "observed_orcids": ["https://orcid.org/0000-0002-8885-0872"]}, "institutions": [{"id": "https://openalex.org/I189196454", "display_name": "The University of Texas at Arlington", "ror": "https://ror.org/019kgqr73", "country_code": "US", "type": "education", "lineage": ["https://openalex.org/I189196454"]}], "countries": ["US"], "is_corresponding": false, "raw_author_name": "Somin Park", "raw_affiliation_strings": ["University of Texas at Arlington, 415 S West St Apt no 205, Arlington, TX TEXAS 76019, United States"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "University of Texas at Arlington, 415 S West St Apt no 205, Arlington, TX TEXAS 76019, United States", "institution_ids": ["https://openalex.org/I189196454"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5100773581", "display_name": "Daeho Kim", "orcid": "https://orcid.org/0000-0003-2921-3168", "observed_orcids": ["https://orcid.org/0000-0003-2921-3168"]}, "institutions": [{"id": "https://openalex.org/I185261750", "display_name": "University of Toronto", "ror": "https://ror.org/03dbr7087", "country_code": "CA", "type": "education", "lineage": ["https://openalex.org/I185261750"]}], "countries": ["CA"], "is_corresponding": false, "raw_author_name": "Daeho Kim", "raw_affiliation_strings": ["University of Toronto, 105 St George Street, Toronto, M5S 3G8, Canada"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "University of Toronto, 105 St George Street, Toronto, M5S 3G8, Canada", "institution_ids": ["https://openalex.org/I185261750"]}]}, {"author_position": "last", "author": {"id": "https://openalex.org/A5089943566", "display_name": "Hongjo Kim", "orcid": "https://orcid.org/0000-0002-2374-3785", "observed_orcids": ["https://orcid.org/0000-0002-2374-3785"]}, "institutions": [{"id": "https://openalex.org/I193775966", "display_name": "Yonsei University", "ror": "https://ror.org/01wjejq96", "country_code": "KR", "type": "education", "lineage": ["https://openalex.org/I193775966"]}], "countries": ["KR"], "is_corresponding": true, "raw_author_name": "Hongjo Kim", "raw_affiliation_strings": ["Yonsei University, Seoul, Korea, Republic of (South Korea)"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Yonsei University, Seoul, Korea, Republic of (South Korea)", "institution_ids": ["https://openalex.org/I193775966"]}]}], "abstract_inverted_index": null, "primary_topic": {"id": "https://openalex.org/T11714", "display_name": "Multimodal Machine Learning Applications", "score": 0.9042999744415283, "subfield": {"id": "https://openalex.org/subfields/1707", "display_name": "Computer Vision and Pattern Recognition"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T11714", "display_name": "Multimodal Machine Learning Applications", "score": 0.9042999744415283, "subfield": {"id": "https://openalex.org/subfields/1707", "display_name": "Computer Vision and Pattern Recognition"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10809", "display_name": "Occupational Health and Safety Research", "score": 0.03830000013113022, "subfield": {"id": "https://openalex.org/subfields/3614", "display_name": "Radiological and Ultrasound Technology"}, "field": {"id": "https://openalex.org/fields/36", "display_name": "Health Professions"}, "domain": {"id": "https://openalex.org/domains/4", "display_name": "Health Sciences"}}, {"id": "https://openalex.org/T10036", "display_name": "Advanced Neural Network Applications", "score": 0.03669999912381172, "subfield": {"id": "https://openalex.org/subfields/1707", "display_name": "Computer Vision and Pattern Recognition"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "doi:10.2139/ssrn.5200113", "is_oa": true, "landing_page_url": "https://doi.org/10.2139/ssrn.5200113", "pdf_url": null, "source": {"id": "https://openalex.org/S4210172589", "display_name": "SSRN Electronic Journal", "issn_l": "1556-5068", "issn": ["1556-5068"], "is_oa": true, "is_in_doaj": false, "is_core": false, "listed_in": [], "host_organization": "https://openalex.org/I1318003438", "host_organization_name": "RELX Group (Netherlands)", "host_organization_lineage": ["https://openalex.org/I1318003438"], "host_organization_lineage_names": [], "type": "repository"}, "license": null, "license_id": null, "version": "acceptedVersion", "is_accepted": true, "is_published": false, "raw_source_name": null, "raw_type": "posted-content"}, "cited_by_count": 0, "is_retracted": false}, {"id": "https://openalex.org/W7126175111", "doi": "https://doi.org/10.18280/isi.301215", "title": "LLM-Cardio: A Large Language Model-Based Assistant for Cardiovascular Health Inquiry and Diagnostic Support Using Wearable Data", "publication_year": 2025, "publication_date": "2025-12-31", "type": "article", "language": null, "authorships": [{"author_position": "first", "author": {"id": "https://openalex.org/A5082972248", "display_name": "Sabrina Mehdi", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Sabrina Mehdi", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5081167604", "display_name": "Sofia Kouah", "orcid": "https://orcid.org/0000-0002-1538-5543", "observed_orcids": ["https://orcid.org/0000-0002-1538-5543"]}, "institutions": [{"id": "https://openalex.org/I3132180716", "display_name": "Larbi Ben M'hidi University of Oum El Bouaghi", "ror": "https://ror.org/0034tbg85", "country_code": "DZ", "type": "education", "lineage": ["https://openalex.org/I3132180716"]}], "countries": ["DZ"], "is_corresponding": false, "raw_author_name": "Sofia Kouah", "raw_affiliation_strings": ["Artificial Intelligence and Autonomous Things Laboratory (IAOA) , University of Oum El Bouaghi , Oum El Bouaghi 04000 , Algeria", "Department of Mathematics and Computer Sciences , University of Oum El Bouaghi , Oum El Bouaghi 04000 , Algeria"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Artificial Intelligence and Autonomous Things Laboratory (IAOA) , University of Oum El Bouaghi , Oum El Bouaghi 04000 , Algeria", "institution_ids": ["https://openalex.org/I3132180716"]}, {"raw_affiliation_string": "Department of Mathematics and Computer Sciences , University of Oum El Bouaghi , Oum El Bouaghi 04000 , Algeria", "institution_ids": ["https://openalex.org/I3132180716"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5007627975", "display_name": "Asma Saighi", "orcid": "https://orcid.org/0000-0003-3790-2262", "observed_orcids": ["https://orcid.org/0000-0003-3790-2262"]}, "institutions": [{"id": "https://openalex.org/I3132180716", "display_name": "Larbi Ben M'hidi University of Oum El Bouaghi", "ror": "https://ror.org/0034tbg85", "country_code": "DZ", "type": "education", "lineage": ["https://openalex.org/I3132180716"]}], "countries": ["DZ"], "is_corresponding": false, "raw_author_name": "Asma Saighi", "raw_affiliation_strings": ["Artificial Intelligence and Autonomous Things Laboratory (IAOA) , University of Oum El Bouaghi , Oum El Bouaghi 04000 , Algeria", "Department of Mathematics and Computer Sciences , University of Oum El Bouaghi , Oum El Bouaghi 04000 , Algeria"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Artificial Intelligence and Autonomous Things Laboratory (IAOA) , University of Oum El Bouaghi , Oum El Bouaghi 04000 , Algeria", "institution_ids": ["https://openalex.org/I3132180716"]}, {"raw_affiliation_string": "Department of Mathematics and Computer Sciences , University of Oum El Bouaghi , Oum El Bouaghi 04000 , Algeria", "institution_ids": ["https://openalex.org/I3132180716"]}]}, {"author_position": "last", "author": {"id": "https://openalex.org/A5072472908", "display_name": "Soumia Zertal", "orcid": "https://orcid.org/0000-0003-0127-900X", "observed_orcids": ["https://orcid.org/0000-0003-0127-900X"]}, "institutions": [{"id": "https://openalex.org/I3132180716", "display_name": "Larbi Ben M'hidi University of Oum El Bouaghi", "ror": "https://ror.org/0034tbg85", "country_code": "DZ", "type": "education", "lineage": ["https://openalex.org/I3132180716"]}], "countries": ["DZ"], "is_corresponding": false, "raw_author_name": "Soumia Zertal", "raw_affiliation_strings": ["Artificial Intelligence and Autonomous Things Laboratory (IAOA) , University of Oum El Bouaghi , Oum El Bouaghi 04000 , Algeria", "Department of Mathematics and Computer Sciences , University of Oum El Bouaghi , Oum El Bouaghi 04000 , Algeria"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Artificial Intelligence and Autonomous Things Laboratory (IAOA) , University of Oum El Bouaghi , Oum El Bouaghi 04000 , Algeria", "institution_ids": ["https://openalex.org/I3132180716"]}, {"raw_affiliation_string": "Department of Mathematics and Computer Sciences , University of Oum El Bouaghi , Oum El Bouaghi 04000 , Algeria", "institution_ids": ["https://openalex.org/I3132180716"]}]}], "abstract_inverted_index": {"The": [0], "global": [1], "increase": [2], "in": [3, 94, 161], "cardiovascular": [4, 55, 104, 199], "disease": [5], "(CVD)": [6], "cases,": [7, 83, 123], "along": [8], "with": [9, 47, 97, 175, 191], "the": [10, 61, 66, 110, 152, 169, 186], "growing": [11], "use": [12], "of": [13, 31, 112, 171], "wearable": [14, 43, 173], "health": [15], "technologies,": [16], "has": [17], "created": [18], "a": [19, 72, 113, 147], "demand": [20], "for": [21, 53, 178, 188], "intelligent": [22, 179], "tools": [23], "that": [24, 41, 75], "support": [25, 28], "early": [26], "diagnostic": [27, 80, 105, 120], "and": [29, 44, 84, 124, 135, 182, 184, 194], "monitoring": [30, 143, 181], "heart": [32], "conditions.This": [33], "work": [34], "introduces": [35], "LLM-Cardio,": [36], "an": [37, 98], "AI-driven": [38], "cardiology": [39], "assistant": [40], "combines": [42], "clinical": [45, 82, 122, 195], "data": [46, 92], "large": [48], "language": [49], "model": [50, 154], "(LLM)": [51], "reasoning": [52, 177], "personalized": [54], "assessment.The": [56], "system": [57, 88], "is": [58, 109], "powered": [59], "by": [60], "Meta-Llama-3.1-8B-Instructmodel": [62], "(4-bit),": [63], "fine-tuned": [64, 153], "using": [65], "LoRA": [67], "(Low-Rank": [68], "Adaptation)": [69], "method": [70], "on": [71, 116, 164], "cardiology-specific": [73, 117], "dataset": [74], "includes": [76], "structured": [77], "medical": [78, 85, 125], "records,": [79], "reports,": [81, 121], "Q&A": [86, 126], "data.The": [87], "integrates": [89], "streaming": [90], "vital-sign": [91], "(simulated": [93], "this": [95], "study)": [96], "instruction-tuned": [99], "LLM": [100, 115], "to": [101], "deliver": [102], "adaptive": [103], "support.A": [106], "key": [107], "contribution": [108], "fine-tuning": [111], "pretrained": [114], "datasets,": [118], "including": [119], "data.Users": [127], "can": [128], "describe": [129], "symptoms": [130], "or": [131], "ask": [132], "cardiology-related": [133], "questions": [134], "receive": [136], "medically": [137], "grounded,": [138], "explainable": [139], "responses,": [140], "while": [141], "simultaneously": [142], "vital": [144], "signs": [145], "through": [146], "responsive": [148], "mobile": [149], "interface.Using": [150], "BERTScore,": [151], "achieved": [155], "Precision=0.9463,Recall=0.9527,F1-score=0.9493,": [156], "outperforming": [157], "baseline": [158], "generative": [159], "models": [160], "semantic": [162], "similarity": [163], "our": [165], "test": [166], "set.LLM-Cardio": [167], "illustrates": [168], "potential": [170], "merging": [172], "technologies": [174], "AI": [176], "cardiac": [180], "diagnosis,": [183], "sets": [185], "groundwork": [187], "future": [189], "integration": [190], "real": [192], "devices": [193], "validation": [196], "toward": [197], "proactive": [198], "care.": [200]}, "primary_topic": {"id": "https://openalex.org/T11636", "display_name": "Artificial Intelligence in Healthcare and Education", "score": 0.8008000254631042, "subfield": {"id": "https://openalex.org/subfields/2718", "display_name": "Health Informatics"}, "field": {"id": "https://openalex.org/fields/27", "display_name": "Medicine"}, "domain": {"id": "https://openalex.org/domains/4", "display_name": "Health Sciences"}}, "topics": [{"id": "https://openalex.org/T11636", "display_name": "Artificial Intelligence in Healthcare and Education", "score": 0.8008000254631042, "subfield": {"id": "https://openalex.org/subfields/2718", "display_name": "Health Informatics"}, "field": {"id": "https://openalex.org/fields/27", "display_name": "Medicine"}, "domain": {"id": "https://openalex.org/domains/4", "display_name": "Health Sciences"}}, {"id": "https://openalex.org/T11396", "display_name": "Artificial Intelligence in Healthcare", "score": 0.17790000140666962, "subfield": {"id": "https://openalex.org/subfields/3605", "display_name": "Health Information Management"}, "field": {"id": "https://openalex.org/fields/36", "display_name": "Health Professions"}, "domain": {"id": "https://openalex.org/domains/4", "display_name": "Health Sciences"}}, {"id": "https://openalex.org/T13702", "display_name": "Machine Learning in Healthcare", "score": 0.006500000134110451, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "doi:10.18280/isi.301215", "is_oa": true, "landing_page_url": "https://doi.org/10.18280/isi.301215", "pdf_url": "https://www.iieta.org/download/file/fid/192156", "source": {"id": "https://openalex.org/S1757003", "display_name": "Ing\u00e9nierie des syst\u00e8mes d information", "issn_l": "1633-1311", "issn": ["1633-1311", "2116-7125"], "is_oa": false, "is_in_doaj": false, "is_core": true, "listed_in": ["cwts-core", "jufo-1", "norway-1"], "host_organization": "https://openalex.org/P4310311056", "host_organization_name": "Lavoisier publishing", "host_organization_lineage": ["https://openalex.org/P4310311056"], "host_organization_lineage_names": ["Lavoisier publishing"], "type": "journal"}, "license": "cc-by", "license_id": "https://openalex.org/licenses/cc-by", "version": "publishedVersion", "is_accepted": true, "is_published": true, "raw_source_name": "Ing\u00e9nierie des syst\u00e8mes d information", "raw_type": "journal-article"}, "cited_by_count": 0, "is_retracted": false}, {"id": "https://openalex.org/W7117761599", "doi": "https://doi.org/10.48550/arxiv.2512.22628", "title": "M2G-Eval: Enhancing and Evaluating Multi-granularity Multilingual Code Generation", "publication_year": 2025, "publication_date": "2025-12-27", "type": "preprint", "language": null, "authorships": [{"author_position": "first", "author": {"id": "https://openalex.org/A5121664679", "display_name": "Fanglin Xu", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Xu, Fanglin", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5121606180", "display_name": "Wei Zhang", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Zhang, Wei", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5121666933", "display_name": "Jian Yang", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Yang, Jian", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5121650985", "display_name": "Guo Chen", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Chen, Guo", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5014870180", "display_name": "Aishan Liu", "orcid": "https://orcid.org/0000-0002-4224-1318", "observed_orcids": ["https://orcid.org/0000-0002-4224-1318"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Liu, Aishan", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5022818809", "display_name": "Zhoujun Li", "orcid": "https://orcid.org/0000-0002-9603-9713", "observed_orcids": ["https://orcid.org/0000-0002-9603-9713"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Li, Zhoujun", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5129944627", "display_name": "Xianglong Liu", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Liu, Xianglong", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "last", "author": {"id": "https://openalex.org/A5121665631", "display_name": "Bryan Dai", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Dai, Bryan", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}], "abstract_inverted_index": {"The": [0], "rapid": [1], "advancement": [2], "of": [3, 160], "code": [4, 18, 44, 58, 161], "large": [5, 61], "language": [6, 62], "models": [7, 26, 63, 91, 105, 151], "(LLMs)": [8, 64], "has": [9], "sparked": [10], "significant": [11], "research": [12], "interest": [13], "in": [14, 60, 168], "systematically": [15], "evaluating": [16, 57], "their": [17], "generation": [19, 59, 162], "capabilities,": [20], "yet": [21], "existing": [22], "benchmarks": [23], "predominantly": [24], "assess": [25], "at": [27], "a": [28, 52], "single": [29], "structural": [30], "granularity": [31], "and": [32, 46, 71, 82, 98, 127, 137, 144, 164], "focus": [33], "on": [34], "limited": [35], "programming": [36, 75, 154], "languages,": [37, 76], "obscuring": [38], "fine-grained": [39, 158], "capability": [40], "variations": [41], "across": [42, 65], "different": [43], "scopes": [45], "multilingual": [47, 54], "scenarios.": [48], "We": [49, 88], "introduce": [50], "M2G-Eval,": [51], "multi-granularity,": [53], "framework": [55], "for": [56], "four": [66], "levels:": [67], "Class,": [68], "Function,": [69], "Block,": [70], "Line.": [72], "Spanning": [73], "18": [74], "M2G-Eval": [77, 156], "includes": [78], "17K+": [79], "training": [80, 93], "tasks": [81, 125], "1,286": [83], "human-annotated,": [84], "contamination-controlled": [85], "test": [86], "instances.": [87], "develop": [89], "M2G-Eval-Coder": [90, 112], "by": [92], "Qwen3-8B": [94], "with": [95, 123], "supervised": [96], "fine-tuning": [97], "Group": [99], "Relative": [100], "Policy": [101], "Optimization.": [102], "Evaluating": [103], "30": [104], "(28": [106], "state-of-the-art": [107], "LLMs": [108], "plus": [109], "our": [110], "two": [111], "variants)": [113], "reveals": [114], "three": [115], "main": [116], "findings:": [117], "(1)": [118], "an": [119], "apparent": [120], "difficulty": [121], "hierarchy,": [122], "Line-level": [124], "easiest": [126], "Class-level": [128], "most": [129], "challenging;": [130], "(2)": [131], "widening": [132], "performance": [133], "gaps": [134], "between": [135], "full-": [136], "partial-granularity": [138], "languages": [139], "as": [140], "task": [141], "complexity": [142], "increases;": [143], "(3)": [145], "strong": [146], "cross-language": [147], "correlations,": [148], "suggesting": [149], "that": [150], "learn": [152], "transferable": [153], "concepts.": [155], "enables": [157], "diagnosis": [159], "capabilities": [163], "highlights": [165], "persistent": [166], "challenges": [167], "synthesizing": [169], "complex,": [170], "long-form": [171], "code.": [172]}, "primary_topic": {"id": "https://openalex.org/T10260", "display_name": "Software Engineering Research", "score": 0.9961000084877014, "subfield": {"id": "https://openalex.org/subfields/1710", "display_name": "Information Systems"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T10260", "display_name": "Software Engineering Research", "score": 0.9961000084877014, "subfield": {"id": "https://openalex.org/subfields/1710", "display_name": "Information Systems"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.0013000000035390258, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10533", "display_name": "Teaching and Learning Programming", "score": 0.000699999975040555, "subfield": {"id": "https://openalex.org/subfields/1706", "display_name": "Computer Science Applications"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "pmh:oai:arXiv.org:2512.22628", "is_oa": true, "landing_page_url": "http://arxiv.org/abs/2512.22628", "pdf_url": "https://arxiv.org/pdf/2512.22628", "source": {"id": "https://openalex.org/S4306400194", "display_name": "arXiv (Cornell University)", "issn_l": "2331-8422", "issn": ["2331-8422"], "is_oa": true, "is_in_doaj": false, "is_core": false, "listed_in": [], "host_organization": "https://openalex.org/I205783295", "host_organization_name": "Cornell University", "host_organization_lineage": ["https://openalex.org/I205783295"], "host_organization_lineage_names": [], "type": "repository"}, "license": null, "license_id": null, "version": "submittedVersion", "is_accepted": false, "is_published": false, "raw_source_name": null, "raw_type": "text"}, "cited_by_count": 0, "is_retracted": false}, {"id": "https://openalex.org/W4415130110", "doi": "https://doi.org/10.48550/arxiv.2506.02018", "title": "Enhancing Paraphrase Type Generation: The Impact of DPO and RLHF Evaluated with Human-Ranked Data", "publication_year": 2025, "publication_date": "2025-05-28", "type": "preprint", "language": "en", "authorships": [{"author_position": "first", "author": {"id": "https://openalex.org/A5119977587", "display_name": "Christopher Lee L\u00fcbbers", "orcid": null, "observed_orcids": []}, "institutions": [{"id": "https://openalex.org/I74656192", "display_name": "University of G\u00f6ttingen", "ror": "https://ror.org/01y9bpm73", "country_code": "DE", "type": "education", "lineage": ["https://openalex.org/I74656192"]}], "countries": ["DE"], "is_corresponding": true, "raw_author_name": "L\u00fcbbers, Christopher Lee", "raw_affiliation_strings": ["University of G\u00f6ttingen , Germany"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "University of G\u00f6ttingen , Germany", "institution_ids": ["https://openalex.org/I74656192"]}]}], "abstract_inverted_index": {"Paraphrasing": [0], "re-expresses": [1], "meaning": [2], "to": [3, 35, 71], "enhance": [4], "applications": [5, 151], "like": [6], "text": [7], "simplification,": [8], "machine": [9], "translation,": [10], "and": [11, 20, 40, 51, 65, 93, 129, 140, 156, 166, 184], "question-answering.": [12, 159], "Specific": [13], "paraphrase": [14, 174], "types": [15], "facilitate": [16], "accurate": [17, 147], "semantic": [18, 49], "analysis": [19], "robust": [21, 158], "language": [22, 182], "models.": [23], "However,": [24], "existing": [25], "paraphrase-type": [26, 63, 82, 114, 177], "generation": [27, 83, 183], "methods": [28], "often": [29], "misalign": [30], "with": [31, 76], "human": [32, 77, 95], "preferences": [33], "due": [34], "reliance": [36], "on": [37], "automated": [38, 164], "metrics": [39, 165], "limited": [41], "human-annotated": [42, 105], "training": [43, 80, 142], "data,": [44], "obscuring": [45], "crucial": [46], "aspects": [47], "of": [48, 120], "fidelity": [50], "linguistic": [52], "transformations.": [53], "This": [54], "study": [55], "addresses": [56], "this": [57], "gap": [58], "by": [59, 85, 98], "leveraging": [60], "a": [61, 90, 113, 168, 186], "human-ranked": [62], "dataset": [64, 106], "integrating": [66], "Direct": [67], "Preference": [68], "Optimization": [69], "(DPO)": [70], "align": [72], "model": [73, 116, 162], "outputs": [74], "directly": [75], "judgments.": [78], "DPO-based": [79], "increases": [81], "accuracy": [84], "3": [86], "percentage": [87, 100], "points": [88], "over": [89], "supervised": [91], "baseline": [92], "raises": [94], "preference": [96, 138], "ratings": [97], "7": [99], "points.": [101], "A": [102], "newly": [103], "created": [104], "supports": [107], "more": [108, 144, 157, 169], "rigorous": [109], "future": [110, 190], "evaluations.": [111], "Additionally,": [112], "detection": [115], "achieves": [117], "F1": [118], "scores": [119], "0.91": [121], "for": [122, 125, 131, 172, 189], "addition/deletion,": [123], "0.78": [124], "same": [126], "polarity": [127], "substitution,": [128], "0.70": [130], "punctuation": [132], "changes.": [133], "These": [134], "findings": [135], "demonstrate": [136], "that": [137], "data": [139], "DPO": [141], "produce": [143], "reliable,": [145], "semantically": [146], "paraphrases,": [148], "enabling": [149], "downstream": [150], "such": [152], "as": [153], "improved": [154], "summarization": [155], "The": [160], "PTD": [161], "surpasses": [163], "provides": [167], "reliable": [170], "framework": [171], "evaluating": [173], "quality,": [175], "advancing": [176], "research": [178], "toward": [179], "richer,": [180], "user-aligned": [181], "establishing": [185], "stronger": [187], "foundation": [188], "evaluations": [191], "grounded": [192], "in": [193], "human-centric": [194], "criteria.": [195]}, "primary_topic": {"id": "https://openalex.org/T10181", "display_name": "Natural Language Processing Techniques", "score": 0.6875, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T10181", "display_name": "Natural Language Processing Techniques", "score": 0.6875, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.28760001063346863, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T13629", "display_name": "Text Readability and Simplification", "score": 0.00570000009611249, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "pmh:oai:arXiv.org:2506.02018", "is_oa": true, "landing_page_url": "https://arxiv.org/abs/2506.02018", "pdf_url": "https://arxiv.org/pdf/2506.02018", "source": {"id": "https://openalex.org/S4306400194", "display_name": "arXiv (Cornell University)", "issn_l": "2331-8422", "issn": ["2331-8422"], "is_oa": true, "is_in_doaj": false, "is_core": false, "listed_in": [], "host_organization": "https://openalex.org/I205783295", "host_organization_name": "Cornell University", "host_organization_lineage": ["https://openalex.org/I205783295"], "host_organization_lineage_names": [], "type": "repository"}, "license": "cc-by-sa", "license_id": "https://openalex.org/licenses/cc-by-sa", "version": "submittedVersion", "is_accepted": false, "is_published": false, "raw_source_name": null, "raw_type": "text"}, "cited_by_count": 0, "is_retracted": false}, {"id": "https://openalex.org/W4413922724", "doi": "https://doi.org/10.1016/j.eswa.2025.129493", "title": "Automating construction contract question answering using large language model and fine-tuning", "publication_year": 2025, "publication_date": "2025-09-02", "type": "article", "language": "en", "authorships": [{"author_position": "first", "author": {"id": null, "display_name": "Mingyu Zhang", "orcid": "https://orcid.org/0009-0009-3883-061X", "observed_orcids": ["https://orcid.org/0009-0009-3883-061X"]}, "institutions": [{"id": "https://openalex.org/I162868743", "display_name": "Tianjin University", "ror": "https://ror.org/012tb2g32", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I162868743"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Mingyu Zhang", "raw_affiliation_strings": ["College of Management and Economics, Tianjin University, China"], "raw_orcid": "https://orcid.org/0009-0009-3883-061X", "affiliations": [{"raw_affiliation_string": "College of Management and Economics, Tianjin University, China", "institution_ids": ["https://openalex.org/I162868743"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5045582317", "display_name": "Chenglong Xu", "orcid": "https://orcid.org/0000-0003-4459-1845", "observed_orcids": ["https://orcid.org/0000-0003-4459-1845"]}, "institutions": [{"id": "https://openalex.org/I200769079", "display_name": "Hong Kong University of Science and Technology", "ror": "https://ror.org/00q4vv597", "country_code": "HK", "type": "education", "lineage": ["https://openalex.org/I200769079"]}], "countries": ["HK"], "is_corresponding": false, "raw_author_name": "Chenglong Xu", "raw_affiliation_strings": ["Department of Civil and Environmental Engineering, Hong Kong University of Science and Technology, Hong Kong"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Department of Civil and Environmental Engineering, Hong Kong University of Science and Technology, Hong Kong", "institution_ids": ["https://openalex.org/I200769079"]}]}, {"author_position": "middle", "author": {"id": null, "display_name": "Yihong Gan", "orcid": null, "observed_orcids": []}, "institutions": [{"id": "https://openalex.org/I162868743", "display_name": "Tianjin University", "ror": "https://ror.org/012tb2g32", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I162868743"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Yihong Gan", "raw_affiliation_strings": ["College of Management and Economics, Tianjin University, China"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "College of Management and Economics, Tianjin University, China", "institution_ids": ["https://openalex.org/I162868743"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5046701060", "display_name": "Yu Wang", "orcid": "https://orcid.org/0000-0002-8768-1093", "observed_orcids": ["https://orcid.org/0000-0002-8768-1093"]}, "institutions": [{"id": "https://openalex.org/I165488957", "display_name": "Lingnan University", "ror": "https://ror.org/0563pg902", "country_code": "HK", "type": "education", "lineage": ["https://openalex.org/I165488957"]}], "countries": ["HK"], "is_corresponding": true, "raw_author_name": "Yu Wang", "raw_affiliation_strings": ["School of Data Science, Lingnan University, Hong Kong"], "raw_orcid": "https://orcid.org/0000-0002-8768-1093", "affiliations": [{"raw_affiliation_string": "School of Data Science, Lingnan University, Hong Kong", "institution_ids": ["https://openalex.org/I165488957"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5101923814", "display_name": "Yi Fu", "orcid": "https://orcid.org/0000-0002-7079-7057", "observed_orcids": ["https://orcid.org/0000-0002-7079-7057", "https://orcid.org/0009-0008-3627-5416"]}, "institutions": [{"id": "https://openalex.org/I162868743", "display_name": "Tianjin University", "ror": "https://ror.org/012tb2g32", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I162868743"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Yi Fu", "raw_affiliation_strings": ["College of Management and Economics, Tianjin University, China"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "College of Management and Economics, Tianjin University, China", "institution_ids": ["https://openalex.org/I162868743"]}]}, {"author_position": "last", "author": {"id": "https://openalex.org/A5024521255", "display_name": "Yongqiang Chen", "orcid": "https://orcid.org/0000-0003-1387-3583", "observed_orcids": ["https://orcid.org/0000-0003-1387-3583"]}, "institutions": [{"id": "https://openalex.org/I162868743", "display_name": "Tianjin University", "ror": "https://ror.org/012tb2g32", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I162868743"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Yongqiang Chen", "raw_affiliation_strings": ["College of Management and Economics, Tianjin University, China"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "College of Management and Economics, Tianjin University, China", "institution_ids": ["https://openalex.org/I162868743"]}]}], "abstract_inverted_index": null, "primary_topic": {"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.9550999999046326, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.9550999999046326, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10181", "display_name": "Natural Language Processing Techniques", "score": 0.014999999664723873, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T13643", "display_name": "Artificial Intelligence in Law", "score": 0.011900000274181366, "subfield": {"id": "https://openalex.org/subfields/3320", "display_name": "Political Science and International Relations"}, "field": {"id": "https://openalex.org/fields/33", "display_name": "Social Sciences"}, "domain": {"id": "https://openalex.org/domains/2", "display_name": "Social Sciences"}}], "primary_location": {"id": "doi:10.1016/j.eswa.2025.129493", "is_oa": false, "landing_page_url": "https://doi.org/10.1016/j.eswa.2025.129493", "pdf_url": null, "source": {"id": "https://openalex.org/S13144211", "display_name": "Expert Systems with Applications", "issn_l": "0957-4174", "issn": ["0957-4174", "1873-6793"], "is_oa": false, "is_in_doaj": false, "is_core": true, "listed_in": ["cwts-core", "doyens", "jufo-2", "ki-jl-2", "norway-2"], "host_organization": "https://openalex.org/P4310320990", "host_organization_name": "Elsevier BV", "host_organization_lineage": ["https://openalex.org/P4310320990"], "host_organization_lineage_names": ["Elsevier BV"], "type": "journal"}, "license": null, "license_id": null, "version": "publishedVersion", "is_accepted": true, "is_published": true, "raw_source_name": "Expert Systems with Applications", "raw_type": "journal-article"}, "cited_by_count": 9, "is_retracted": false}, {"id": "https://openalex.org/W4415333017", "doi": "https://doi.org/10.48550/arxiv.2506.14903", "title": "DETONATE: A Benchmark for Text-to-Image Alignment and Kernelized Direct Preference Optimization", "publication_year": 2025, "publication_date": "2025-06-17", "type": "preprint", "language": "en", "authorships": [{"author_position": "first", "author": {"id": "https://openalex.org/A5113174281", "display_name": "Renjith Prasad", "orcid": "https://orcid.org/0009-0005-0362-5844", "observed_orcids": ["https://orcid.org/0009-0005-0362-5844"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Prasad, Renjith", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": null, "display_name": "Borah, Abhilekh", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Borah, Abhilekh", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5009080427", "display_name": "Hasnat Abdullah", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Abdullah, Hasnat Md", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5079093269", "display_name": "Chathurangi Shyalika", "orcid": "https://orcid.org/0000-0002-5320-5566", "observed_orcids": ["https://orcid.org/0000-0002-5320-5566"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Shyalika, Chathurangi", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": null, "display_name": "Singh, Gurpreet", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Singh, Gurpreet", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": null, "display_name": "Garimella, Ritvik", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Garimella, Ritvik", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5078181730", "display_name": "Rajarshi Roy", "orcid": "https://orcid.org/0000-0002-0202-1169", "observed_orcids": ["https://orcid.org/0000-0002-0202-1169"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Roy, Rajarshi", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5090455695", "display_name": "Harshul Surana", "orcid": "https://orcid.org/0009-0005-9379-1087", "observed_orcids": ["https://orcid.org/0009-0005-9379-1087"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Surana, Harshul", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5056733281", "display_name": "Nasrin Imanpour", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Imanpour, Nasrin", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5017319608", "display_name": "Suranjana Trivedy", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Trivedy, Suranjana", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": null, "display_name": "Sheth, Amit", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Sheth, Amit", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "last", "author": {"id": "https://openalex.org/A5101679308", "display_name": "Amitava Das", "orcid": "https://orcid.org/0000-0003-3418-463X", "observed_orcids": ["https://orcid.org/0000-0003-3418-463X"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Das, Amitava", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}], "abstract_inverted_index": {"Alignment": [0, 174], "is": [1, 31], "crucial": [2], "for": [3, 42, 63, 78, 107], "text-to-image": [4], "(T2I)": [5], "models": [6, 29, 159], "to": [7, 35], "ensure": [8], "that": [9, 195], "generated": [10, 155], "images": [11, 154], "faithfully": [12], "capture": [13], "user": [14], "intent": [15], "while": [16], "maintaining": [17], "safety": [18], "and": [19, 75, 82, 87, 90, 104, 110, 131, 140, 144, 168, 206], "fairness.": [20], "Direct": [21], "Preference": [22], "Optimization": [23], "(DPO),": [24], "prominent": [25], "in": [26], "large": [27], "language": [28], "(LLMs),": [30], "extending": [32], "its": [33, 120], "influence": [34], "T2I": [36, 43, 158], "systems.": [37], "This": [38], "paper": [39], "introduces": [40], "DPO-Kernels": [41, 196], "models,": [44], "a": [45, 178], "novel": [46, 179], "extension": [47], "enhancing": [48], "alignment": [49], "across": [50], "three": [51, 135], "dimensions:": [52], "(i)": [53], "Hybrid": [54], "Loss,": [55], "integrating": [56], "embedding-based": [57], "objectives": [58], "with": [59, 153], "traditional": [60], "probability-based": [61], "loss": [62], "improved": [64], "optimization;": [65], "(ii)": [66], "Kernelized": [67], "Representations,": [68], "employing": [69], "Radial": [70], "Basis": [71], "Function": [72], "(RBF),": [73], "Polynomial,": [74], "Wavelet": [76], "kernels": [77], "richer": [79], "feature": [80], "transformations": [81], "better": [83], "separation": [84], "between": [85], "safe": [86], "unsafe": [88], "inputs;": [89], "(iii)": [91], "Divergence": [92], "Selection,": [93], "expanding": [94], "beyond": [95], "DPO's": [96], "default": [97], "Kullback-Leibler": [98], "(KL)": [99], "regularizer": [100], "by": [101, 156], "incorporating": [102], "Wasserstein": [103], "R'enyi": [105], "divergences": [106], "enhanced": [108], "stability": [109], "robustness.": [111], "We": [112], "introduce": [113], "DETONATE,": [114], "the": [115, 173], "first": [116], "large-scale": [117], "benchmark": [118], "of": [119, 137, 185], "kind,": [121], "comprising": [122], "approximately": [123], "100K": [124], "curated": [125], "image": [126, 187], "pairs": [127], "categorized": [128], "as": [129], "chosen": [130], "rejected.": [132], "DETONATE": [133, 205], "encapsulates": [134], "axes": [136], "social": [138], "bias": [139], "discrimination:": [141], "Race,": [142], "Gender,": [143], "Disability.": [145], "Prompts": [146], "are": [147, 209], "sourced": [148], "from": [149], "hate": [150], "speech": [151], "datasets,": [152], "leading": [157], "including": [160], "Stable": [161, 165], "Diffusion": [162, 166], "3.5": [163], "Large,": [164], "XL,": [167], "Midjourney.": [169], "Additionally,": [170], "we": [171, 193], "propose": [172], "Quality": [175], "Index": [176], "(AQI),": [177], "geometric": [180], "measure": [181], "quantifying": [182], "latent-space": [183], "separability": [184], "safe/unsafe": [186], "activations,": [188], "revealing": [189], "hidden": [190], "vulnerabilities.": [191], "Empirically,": [192], "demonstrate": [194], "maintain": [197], "strong": [198], "generalization": [199], "bounds": [200], "via": [201], "Heavy-Tailed": [202], "Self-Regularization": [203], "(HT-SR).": [204], "complete": [207], "code": [208], "publicly": [210], "released.": [211]}, "primary_topic": {"id": "https://openalex.org/T10775", "display_name": "Generative Adversarial Networks and Image Synthesis", "score": 0.7440999746322632, "subfield": {"id": "https://openalex.org/subfields/1707", "display_name": "Computer Vision and Pattern Recognition"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T10775", "display_name": "Generative Adversarial Networks and Image Synthesis", "score": 0.7440999746322632, "subfield": {"id": "https://openalex.org/subfields/1707", "display_name": "Computer Vision and Pattern Recognition"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10883", "display_name": "Ethics and Social Impacts of AI", "score": 0.12120000272989273, "subfield": {"id": "https://openalex.org/subfields/3311", "display_name": "Safety Research"}, "field": {"id": "https://openalex.org/fields/33", "display_name": "Social Sciences"}, "domain": {"id": "https://openalex.org/domains/2", "display_name": "Social Sciences"}}, {"id": "https://openalex.org/T11689", "display_name": "Adversarial Robustness in Machine Learning", "score": 0.050700001418590546, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "pmh:oai:arXiv.org:2506.14903", "is_oa": true, "landing_page_url": "http://arxiv.org/abs/2506.14903", "pdf_url": "https://arxiv.org/pdf/2506.14903", "source": {"id": "https://openalex.org/S4306400194", "display_name": "arXiv (Cornell University)", "issn_l": "2331-8422", "issn": ["2331-8422"], "is_oa": true, "is_in_doaj": false, "is_core": false, "listed_in": [], "host_organization": "https://openalex.org/I205783295", "host_organization_name": "Cornell University", "host_organization_lineage": ["https://openalex.org/I205783295"], "host_organization_lineage_names": [], "type": "repository"}, "license": null, "license_id": null, "version": "submittedVersion", "is_accepted": false, "is_published": false, "raw_source_name": null, "raw_type": "text"}, "cited_by_count": 0, "is_retracted": false}, {"id": "https://openalex.org/W4415084239", "doi": "https://doi.org/10.48550/arxiv.2503.01131", "title": "Beyond QA Pairs: Assessing Parameter-Efficient Fine-Tuning for Fact Embedding in LLMs", "publication_year": 2025, "publication_date": "2025-03-03", "type": "preprint", "language": "en", "authorships": [{"author_position": "first", "author": {"id": "https://openalex.org/A5139725988", "display_name": "Shivam Ratnakar", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Ratnakar, Shivam", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5119957682", "display_name": "Abhiroop Talasila", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Talasila, Abhiroop", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5119957683", "display_name": "Raghav Chamadiya", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Chamadiya, Raghav", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": null, "display_name": "Agarwal, Nikhil", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Agarwal, Nikhil", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "last", "author": {"id": "https://openalex.org/A5119957684", "display_name": "Vinayak K Doifode", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Doifode, Vinayak K", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}], "abstract_inverted_index": {"This": [0], "paper": [1], "presents": [2], "an": [3], "extensive": [4], "examination": [5], "of": [6, 80, 159, 172], "Parameter-Efficient": [7], "Fine-Tuning": [8], "(PEFT)": [9], "for": [10, 112], "embedding": [11, 113], "domain": [12], "specific": [13, 175], "facts": [14, 114], "into": [15, 31, 115], "Large": [16], "Language": [17], "Models": [18], "(LLMs),": [19], "focusing": [20], "on": [21, 47, 66, 72], "improving": [22], "the": [23, 78, 108, 135, 140, 147, 157, 170], "fine-tuning": [24, 83], "process": [25], "by": [26, 130], "categorizing": [27], "question-answer": [28], "(QA)": [29], "pairs": [30], "Factual": [32], "and": [33, 50, 58, 88, 163], "Conceptual": [34], "classes": [35], "using": [36, 52], "a": [37, 131], "BERT-based": [38], "classifier.": [39], "Two": [40], "distinct": [41], "Llama-2": [42, 142], "models": [43, 54, 64], "are": [44, 128], "fine-tuned": [45, 141], "based": [46], "these": [48], "classifications": [49], "evaluated": [51], "larger": [53], "like": [55], "GPT-3.5": [56], "Turbo": [57], "Gemini.": [59], "Our": [60, 126, 154], "results": [61], "indicate": [62], "that": [63, 103], "trained": [65, 71], "conceptual": [67], "datasets": [68], "outperform": [69], "those": [70], "factual": [73], "datasets.": [74], "Additionally,": [75], "we": [76], "compare": [77], "efficiency": [79], "two": [81], "synthetic": [82, 164], "dataset": [84, 133, 165], "generation": [85, 166], "techniques,": [86], "D-RAG": [87], "D-Naive,": [89], "with": [90], "D-Naive": [91], "demonstrating": [92], "superior": [93], "performance.": [94], "Although": [95], "PEFT": [96], "has": [97, 119], "shown": [98], "effectiveness,": [99], "our": [100], "research": [101], "indicates": [102], "it": [104, 118], "may": [105], "not": [106], "be": [107], "most": [109], "optimal": [110], "method": [111], "LLMs.": [116], "However,": [117], "demonstrated": [120], "exceptional": [121], "performance": [122, 171], "in": [123, 134, 150, 168, 174], "instruction-based": [124], "tasks.": [125], "findings": [127], "reinforced": [129], "1000-sample": [132], "data": [136], "center": [137], "domain,": [138], "where": [139], "7B": [143], "model": [144, 149], "significantly": [145], "outperforms": [146], "baseline": [148], "generating": [151], "product": [152], "recommendations.": [153], "study": [155], "highlights": [156], "importance": [158], "QA": [160], "pair": [161], "categorization": [162], "techniques": [167], "enhancing": [169], "LLMs": [173], "domains.": [176]}, "primary_topic": {"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.9452999830245972, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.9452999830245972, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T11307", "display_name": "Domain Adaptation and Few-Shot Learning", "score": 0.04969999939203262, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T11714", "display_name": "Multimodal Machine Learning Applications", "score": 0.0007999999797903001, "subfield": {"id": "https://openalex.org/subfields/1707", "display_name": "Computer Vision and Pattern Recognition"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "pmh:oai:arXiv.org:2503.01131", "is_oa": true, "landing_page_url": "http://arxiv.org/abs/2503.01131", "pdf_url": "https://arxiv.org/pdf/2503.01131", "source": {"id": "https://openalex.org/S4306400194", "display_name": "arXiv (Cornell University)", "issn_l": "2331-8422", "issn": ["2331-8422"], "is_oa": true, "is_in_doaj": false, "is_core": false, "listed_in": [], "host_organization": "https://openalex.org/I205783295", "host_organization_name": "Cornell University", "host_organization_lineage": ["https://openalex.org/I205783295"], "host_organization_lineage_names": [], "type": "repository"}, "license": null, "license_id": null, "version": "submittedVersion", "is_accepted": false, "is_published": false, "raw_source_name": null, "raw_type": "text"}, "cited_by_count": 0, "is_retracted": false}, {"id": "https://openalex.org/W7130682069", "doi": "https://doi.org/10.5281/zenodo.18712736", "title": "ENHANCING CODE-STYLE RECOGNITION WITH LLMS: DATASET SIZE CONSIDERATIONS", "publication_year": 2025, "publication_date": "2025-11-20", "type": "article", "language": "en", "authorships": [{"author_position": "first", "author": {"id": "https://openalex.org/A5126463051", "display_name": "Mykhailovych Petrenko Oleksandr", "orcid": null, "observed_orcids": []}, "institutions": [{"id": "https://openalex.org/I98435010", "display_name": "Lviv Polytechnic National University", "ror": "https://ror.org/0542q3127", "country_code": "UA", "type": "education", "lineage": ["https://openalex.org/I98435010"]}], "countries": ["UA"], "is_corresponding": true, "raw_author_name": "Oleksandr, Mykhailovych Petrenko", "raw_affiliation_strings": ["Department of Applied Mathematics, Lviv Polytechnic National University, Ukraine"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Department of Applied Mathematics, Lviv Polytechnic National University, Ukraine", "institution_ids": ["https://openalex.org/I98435010"]}]}], "abstract_inverted_index": {"One": [0], "aspect": [1], "of": [2, 44, 56, 69, 118, 144, 176, 199], "a": [3, 10], "well-written": [4], "codebase": [5], "is": [6, 50, 65, 90], "its": [7, 101], "adherence": [8, 184], "to": [9, 28, 60, 75, 105, 124, 146, 172, 185, 194], "particular": [11], "code": [12, 27, 78, 180], "style,": [13], "and": [14, 24, 100, 111, 139, 155, 161, 178, 183], "Large": [15], "Language": [16], "Models": [17], "(LLMs)": [18], "can": [19, 202], "greatly": [20], "assist": [21], "in": [22, 92, 188, 209], "reviewing": [23], "adapting": [25], "the": [26, 30, 42, 45, 53, 85, 107, 132, 174, 196], "follow": [29], "defined": [31], "conventions.": [32], "Because": [33], "specific": [34, 86], "code-style": [35], "rules": [36], "are": [37, 218], "typically": [38], "not": [39], "known": [40], "during": [41], "pre-training": [43], "base": [46], "model,": [47], "additional": [48], "fine-tuning": [49, 73, 207], "necessary.": [51], "However,": [52], "exact": [54], "number": [55, 198], "training": [57, 126, 200], "samples": [58, 201], "required": [59], "achieve": [61], "optimal": [62], "model": [63], "performance": [64], "unclear.": [66], "The": [67, 191], "significance": [68], "dataset": [70], "size": [71], "when": [72], "LLMs": [74, 177, 208], "categorize": [76], "Python": [77], "snippets": [79], "as": [80, 221], "compliant": [81], "or": [82, 215, 228], "noncompliant": [83], "with": [84, 135, 151], "PEP-8": [87], "indentation": [88], "rule": [89], "investigated": [91], "this": [93], "work.": [94], "We": [95], "used": [96, 193], "Low-Rank": [97], "Adaptation": [98], "(LoRA)": [99], "quantized": [102], "variant": [103], "(QLoRA)": [104], "fine-tune": [106], "Llama": [108, 112], "2": [109], "7B": [110], "3": [113], "8B": [114], "models": [115, 133], "on": [116, 169], "datasets": [117, 137, 153], "varying": [119], "sizes,": [120], "ranging": [121], "from": [122], "60": [123], "480": [125, 140], "samples.": [127], "Our": [128], "experiments": [129], "demonstrated": [130], "that": [131], "fine-tuned": [134], "larger": [136], "(240": [138], "samples)": [141, 157], "achieved": [142], "accuracies": [143], "up": [145], "99%,": [147], "whereas": [148], "those": [149], "trained": [150], "smaller": [152], "(60": [154], "120": [156], "experienced": [158], "over": [159], "fitting": [160], "lower": [162], "accuracy.": [163], "Subsequent": [164], "research": [165], "will": [166], "be": [167, 204], "based": [168], "these": [170], "findings": [171], "explore": [173], "potential": [175], "improve": [179], "readability,": [181], "maintainability,": [182], "coding": [186], "standards": [187], "software": [189], "development.": [190], "methodology": [192], "determine": [195], "sufficient": [197], "also": [203], "valuable": [205], "for": [206], "other": [210], "domains": [211], "where": [212], "strict": [213], "style": [214], "formatting": [216], "conventions": [217], "required,": [219], "such": [220], "legal": [222], "document": [223], "preparation,": [224], "standardized": [225], "medical": [226], "reporting,": [227], "financial": [229], "regulatory": [230], "filings.": [231]}, "primary_topic": {"id": "https://openalex.org/T10260", "display_name": "Software Engineering Research", "score": 0.9970999956130981, "subfield": {"id": "https://openalex.org/subfields/1710", "display_name": "Information Systems"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T10260", "display_name": "Software Engineering Research", "score": 0.9970999956130981, "subfield": {"id": "https://openalex.org/subfields/1710", "display_name": "Information Systems"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.0015999999595806003, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10430", "display_name": "Software Engineering Techniques and Practices", "score": 0.00019999999494757503, "subfield": {"id": "https://openalex.org/subfields/1710", "display_name": "Information Systems"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "doi:10.5281/zenodo.18712736", "is_oa": true, "landing_page_url": "https://doi.org/10.5281/zenodo.18712736", "pdf_url": null, "source": {"id": "https://openalex.org/S4306400562", "display_name": "Zenodo (CERN European Organization for Nuclear Research)", "issn_l": null, "issn": null, "is_oa": true, "is_in_doaj": false, "is_core": false, "listed_in": [], "host_organization": "https://openalex.org/I67311998", "host_organization_name": "European Organization for Nuclear Research", "host_organization_lineage": ["https://openalex.org/I67311998"], "host_organization_lineage_names": [], "type": "repository"}, "license": "cc-by", "license_id": "https://openalex.org/licenses/cc-by", "version": null, "is_accepted": false, "is_published": false, "raw_source_name": null, "raw_type": "article-journal"}, "cited_by_count": 0, "is_retracted": false}, {"id": "https://openalex.org/W4415837429", "doi": "https://doi.org/10.21203/rs.3.rs-7864505/v1", "title": "Automated ACR TI-RADS Classification of Thyroid Nodules from Narrative Ultrasound Reports Using a Fine-Tuned Open-Source Language Model: A Reproducible and Low-Resource Framework", "publication_year": 2025, "publication_date": "2025-11-04", "type": "preprint", "language": null, "authorships": [{"author_position": "first", "author": {"id": "https://openalex.org/A5073319456", "display_name": "Miao Yu", "orcid": "https://orcid.org/0000-0002-2804-6014", "observed_orcids": ["https://orcid.org/0000-0002-2804-6014", "https://orcid.org/0000-0002-9080-5931"]}, "institutions": [{"id": "https://openalex.org/I126520041", "display_name": "University of Science and Technology of China", "ror": "https://ror.org/04c4dkn09", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I126520041", "https://openalex.org/I19820366"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Miao Yu", "raw_affiliation_strings": ["The First Affiliated Hospital of University of Science and Technology of China"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "The First Affiliated Hospital of University of Science and Technology of China", "institution_ids": ["https://openalex.org/I126520041"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5101627959", "display_name": "Sijia Huang", "orcid": "https://orcid.org/0009-0003-3927-9176", "observed_orcids": ["https://orcid.org/0009-0003-3927-9176"]}, "institutions": [{"id": "https://openalex.org/I126520041", "display_name": "University of Science and Technology of China", "ror": "https://ror.org/04c4dkn09", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I126520041", "https://openalex.org/I19820366"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Sijia Huang", "raw_affiliation_strings": ["The First Affiliated Hospital of University of Science and Technology of China (Department of Clinical Pathology)"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "The First Affiliated Hospital of University of Science and Technology of China (Department of Clinical Pathology)", "institution_ids": ["https://openalex.org/I126520041"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5111063986", "display_name": "Muyang Li", "orcid": null, "observed_orcids": []}, "institutions": [{"id": "https://openalex.org/I126520041", "display_name": "University of Science and Technology of China", "ror": "https://ror.org/04c4dkn09", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I126520041", "https://openalex.org/I19820366"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Muyang Li", "raw_affiliation_strings": ["The First Affiliated Hospital of University of Science and Technology of China"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "The First Affiliated Hospital of University of Science and Technology of China", "institution_ids": ["https://openalex.org/I126520041"]}]}, {"author_position": "middle", "author": {"id": null, "display_name": "Likuan Zhang", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Likuan Zhang", "raw_affiliation_strings": ["Anqing No. 116 Hospital (Department of Oncology)"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Anqing No. 116 Hospital (Department of Oncology)", "institution_ids": []}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5101616656", "display_name": "Heng Zhang", "orcid": "https://orcid.org/0000-0002-7073-9930", "observed_orcids": ["https://orcid.org/0000-0002-7073-9930"]}, "institutions": [{"id": "https://openalex.org/I126520041", "display_name": "University of Science and Technology of China", "ror": "https://ror.org/04c4dkn09", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I126520041", "https://openalex.org/I19820366"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Heng Zhang", "raw_affiliation_strings": ["The First Affiliated Hospital of University of Science and Technology of China"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "The First Affiliated Hospital of University of Science and Technology of China", "institution_ids": ["https://openalex.org/I126520041"]}]}, {"author_position": "middle", "author": {"id": null, "display_name": "Qiao Xu", "orcid": null, "observed_orcids": []}, "institutions": [{"id": "https://openalex.org/I126520041", "display_name": "University of Science and Technology of China", "ror": "https://ror.org/04c4dkn09", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I126520041", "https://openalex.org/I19820366"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Qiao Xu", "raw_affiliation_strings": ["The First Affiliated Hospital of University of Science and Technology of China"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "The First Affiliated Hospital of University of Science and Technology of China", "institution_ids": ["https://openalex.org/I126520041"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5053965487", "display_name": "Zikang Wang", "orcid": "https://orcid.org/0000-0003-4995-5521", "observed_orcids": ["https://orcid.org/0000-0003-4995-5521"]}, "institutions": [{"id": "https://openalex.org/I126520041", "display_name": "University of Science and Technology of China", "ror": "https://ror.org/04c4dkn09", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I126520041", "https://openalex.org/I19820366"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Zikang Wang", "raw_affiliation_strings": ["The First Affiliated Hospital of University of Science and Technology of China"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "The First Affiliated Hospital of University of Science and Technology of China", "institution_ids": ["https://openalex.org/I126520041"]}]}, {"author_position": "last", "author": {"id": "https://openalex.org/A5101660842", "display_name": "Jian Gao", "orcid": "https://orcid.org/0000-0003-4746-0824", "observed_orcids": ["https://orcid.org/0000-0003-4746-0824"]}, "institutions": [{"id": "https://openalex.org/I126520041", "display_name": "University of Science and Technology of China", "ror": "https://ror.org/04c4dkn09", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I126520041", "https://openalex.org/I19820366"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Jian Gao", "raw_affiliation_strings": ["The First Affiliated Hospital of University of Science and Technology of China"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "The First Affiliated Hospital of University of Science and Technology of China", "institution_ids": ["https://openalex.org/I126520041"]}]}], "abstract_inverted_index": null, "primary_topic": {"id": "https://openalex.org/T10329", "display_name": "Thyroid Cancer Diagnosis and Treatment", "score": 0.3127000033855438, "subfield": {"id": "https://openalex.org/subfields/2712", "display_name": "Endocrinology, Diabetes and Metabolism"}, "field": {"id": "https://openalex.org/fields/27", "display_name": "Medicine"}, "domain": {"id": "https://openalex.org/domains/4", "display_name": "Health Sciences"}}, "topics": [{"id": "https://openalex.org/T10329", "display_name": "Thyroid Cancer Diagnosis and Treatment", "score": 0.3127000033855438, "subfield": {"id": "https://openalex.org/subfields/2712", "display_name": "Endocrinology, Diabetes and Metabolism"}, "field": {"id": "https://openalex.org/fields/27", "display_name": "Medicine"}, "domain": {"id": "https://openalex.org/domains/4", "display_name": "Health Sciences"}}, {"id": "https://openalex.org/T11636", "display_name": "Artificial Intelligence in Healthcare and Education", "score": 0.11599999666213989, "subfield": {"id": "https://openalex.org/subfields/2718", "display_name": "Health Informatics"}, "field": {"id": "https://openalex.org/fields/27", "display_name": "Medicine"}, "domain": {"id": "https://openalex.org/domains/4", "display_name": "Health Sciences"}}, {"id": "https://openalex.org/T13702", "display_name": "Machine Learning in Healthcare", "score": 0.11330000311136246, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "doi:10.21203/rs.3.rs-7864505/v1", "is_oa": true, "landing_page_url": "https://doi.org/10.21203/rs.3.rs-7864505/v1", "pdf_url": "https://www.researchsquare.com/article/rs-7864505/latest.pdf", "source": {"id": "https://openalex.org/S4306525896", "display_name": "Research Square", "issn_l": null, "issn": null, "is_oa": false, "is_in_doaj": false, "is_core": false, "listed_in": [], "host_organization": null, "host_organization_name": null, "host_organization_lineage": [], "host_organization_lineage_names": [], "type": "repository"}, "license": "cc-by", "license_id": "https://openalex.org/licenses/cc-by", "version": "acceptedVersion", "is_accepted": true, "is_published": false, "raw_source_name": "Research Square", "raw_type": "posted-content"}, "cited_by_count": 0, "is_retracted": false}, {"id": "https://openalex.org/W4414487632", "doi": "https://doi.org/10.48550/arxiv.2508.12512", "title": "LangVision-LoRA-NAS: Neural Architecture Search for Variable LoRA Rank in Vision Language Models", "publication_year": 2025, "publication_date": "2025-08-17", "type": "preprint", "language": "en", "authorships": [{"author_position": "first", "author": {"id": "https://openalex.org/A5015607198", "display_name": "Krishna Teja Chitty-Venkata", "orcid": "https://orcid.org/0000-0002-3027-1915", "observed_orcids": ["https://orcid.org/0000-0002-3027-1915"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Chitty-Venkata, Krishna Teja", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5014448094", "display_name": "Murali Krishna Emani", "orcid": "https://orcid.org/0000-0002-6279-0007", "observed_orcids": ["https://orcid.org/0000-0002-6279-0007"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Emani, Murali", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "last", "author": {"id": "https://openalex.org/A5075500139", "display_name": "Venkatram Vishwanath", "orcid": "https://orcid.org/0000-0001-7248-6116", "observed_orcids": ["https://orcid.org/0000-0001-7248-6116"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Vishwanath, Venkatram", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}], "abstract_inverted_index": {"Vision": [0, 20], "Language": [1, 30], "Models": [2], "(VLMs)": [3], "integrate": [4], "visual": [5], "and": [6, 13, 27, 83, 129, 155, 165], "text": [7, 34], "modalities": [8], "to": [9, 44, 48, 55, 103, 113, 123], "enable": [10], "multimodal": [11, 125], "understanding": [12], "generation.": [14, 35], "These": [15], "models": [16, 47, 69, 158], "typically": [17], "combine": [18], "a": [19, 28, 63, 77, 92], "Transformer": [21], "(ViT)": [22], "as": [23, 62], "an": [24, 40], "image": [25], "encoder": [26], "Large": [29], "Model": [31], "(LLM)": [32], "for": [33, 66, 106, 116, 168], "LoRA": [36, 59, 102, 119], "(Low-Rank": [37], "Adaptation)": [38], "is": [39], "efficient": [41], "fine-tuning": [42, 67, 151], "method": [43], "adapt": [45], "pre-trained": [46], "new": [49], "tasks": [50], "by": [51, 70], "introducing": [52, 71], "low-rank": [53, 72], "updates": [54], "their": [56], "weights.": [57], "While": [58], "has": [60], "emerged": [61], "powerful": [64], "technique": [65], "large": [68], "updates,": [73], "current": [74], "implementations": [75], "assume": [76], "fixed": [78], "rank,": [79], "potentially": [80], "limiting": [81], "flexibility": [82], "efficiency": [84], "across": [85], "diverse": [86], "tasks.": [87], "This": [88], "paper": [89], "introduces": [90], "\\textit{LangVision-LoRA-NAS},": [91], "novel": [93], "framework": [94], "that": [95], "integrates": [96], "Neural": [97], "Architecture": [98], "Search": [99], "(NAS)": [100], "with": [101], "optimize": [104], "VLMs": [105], "variable-rank": [107], "adaptation.": [108], "Our": [109, 153], "approach": [110], "leverages": [111], "NAS": [112], "dynamically": [114], "search": [115], "the": [117, 136, 166], "optimal": [118], "rank": [120], "configuration": [121], "tailored": [122], "specific": [124], "tasks,": [126], "balancing": [127], "performance": [128, 148], "computational": [130], "efficiency.": [131], "Through": [132], "extensive": [133], "experiments": [134], "using": [135], "LLaMA-3.2-11B": [137], "model": [138, 147], "on": [139, 159], "several": [140], "datasets,": [141], "LangVision-LoRA-NAS": [142, 169], "demonstrates": [143], "notable": [144], "improvement": [145], "in": [146], "while": [149], "reducing": [150], "costs.": [152], "Base": [154], "searched": [156], "fine-tuned": [157], "LLaMA-3.2-11B-Vision-Instruct": [160], "can": [161, 170], "be": [162, 171], "found": [163, 172], "\\href{https://huggingface.co/collections/krishnateja95/llama-32-11b-vision-instruct-langvision-lora-nas-6786cac480357a6a6fcc59ee}{\\textcolor{blue}{here}}": [164], "code": [167], "\\href{https://github.com/krishnateja95/LangVision-NAS}{\\textcolor{blue}{here}}.": [173]}, "primary_topic": {"id": "https://openalex.org/T11714", "display_name": "Multimodal Machine Learning Applications", "score": 0.9667999744415283, "subfield": {"id": "https://openalex.org/subfields/1707", "display_name": "Computer Vision and Pattern Recognition"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T11714", "display_name": "Multimodal Machine Learning Applications", "score": 0.9667999744415283, "subfield": {"id": "https://openalex.org/subfields/1707", "display_name": "Computer Vision and Pattern Recognition"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T11307", "display_name": "Domain Adaptation and Few-Shot Learning", "score": 0.016899999231100082, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10036", "display_name": "Advanced Neural Network Applications", "score": 0.012600000016391277, "subfield": {"id": "https://openalex.org/subfields/1707", "display_name": "Computer Vision and Pattern Recognition"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "pmh:oai:arXiv.org:2508.12512", "is_oa": true, "landing_page_url": "http://arxiv.org/abs/2508.12512", "pdf_url": "https://arxiv.org/pdf/2508.12512", "source": {"id": "https://openalex.org/S4306400194", "display_name": "arXiv (Cornell University)", "issn_l": "2331-8422", "issn": ["2331-8422"], "is_oa": true, "is_in_doaj": false, "is_core": false, "listed_in": [], "host_organization": "https://openalex.org/I205783295", "host_organization_name": "Cornell University", "host_organization_lineage": ["https://openalex.org/I205783295"], "host_organization_lineage_names": [], "type": "repository"}, "license": null, "license_id": null, "version": "submittedVersion", "is_accepted": false, "is_published": false, "raw_source_name": null, "raw_type": "text"}, "cited_by_count": 0, "is_retracted": false}, {"id": "https://openalex.org/W4417302269", "doi": "https://doi.org/10.48550/arxiv.2505.12871", "title": "Does Low Rank Adaptation Lead to Lower Robustness against Training-Time Attacks?", "publication_year": 2025, "publication_date": "2025-05-19", "type": "preprint", "language": "en", "authorships": [{"author_position": "first", "author": {"id": "https://openalex.org/A5104035702", "display_name": "Zi Liang", "orcid": "https://orcid.org/0009-0003-1418-9537", "observed_orcids": ["https://orcid.org/0009-0003-1418-9537", "https://orcid.org/0009-0001-2265-5388"]}, "institutions": [{"id": "https://openalex.org/I14243506", "display_name": "Hong Kong Polytechnic University", "ror": "https://ror.org/0030zas98", "country_code": "HK", "type": "education", "lineage": ["https://openalex.org/I14243506"]}], "countries": ["HK"], "is_corresponding": false, "raw_author_name": "Liang, Zi", "raw_affiliation_strings": ["The Hong Kong Polytechnic University , Hong Kong , China"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "The Hong Kong Polytechnic University , Hong Kong , China", "institution_ids": ["https://openalex.org/I14243506"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5051088560", "display_name": "Haibo Hu", "orcid": "https://orcid.org/0000-0002-9008-2112", "observed_orcids": ["https://orcid.org/0000-0002-9008-2112"]}, "institutions": [{"id": "https://openalex.org/I14243506", "display_name": "Hong Kong Polytechnic University", "ror": "https://ror.org/0030zas98", "country_code": "HK", "type": "education", "lineage": ["https://openalex.org/I14243506"]}], "countries": ["HK"], "is_corresponding": false, "raw_author_name": "Hu, Haibo", "raw_affiliation_strings": ["The Hong Kong Polytechnic University , Hong Kong , China"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "The Hong Kong Polytechnic University , Hong Kong , China", "institution_ids": ["https://openalex.org/I14243506"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5026389260", "display_name": "Qingqing Ye", "orcid": "https://orcid.org/0000-0003-1547-2847", "observed_orcids": ["https://orcid.org/0000-0003-1547-2847"]}, "institutions": [{"id": "https://openalex.org/I14243506", "display_name": "Hong Kong Polytechnic University", "ror": "https://ror.org/0030zas98", "country_code": "HK", "type": "education", "lineage": ["https://openalex.org/I14243506"]}], "countries": ["HK"], "is_corresponding": false, "raw_author_name": "Ye, Qingqing", "raw_affiliation_strings": ["The Hong Kong Polytechnic University , Hong Kong , China"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "The Hong Kong Polytechnic University , Hong Kong , China", "institution_ids": ["https://openalex.org/I14243506"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5109796060", "display_name": "Yaxin Xiao", "orcid": null, "observed_orcids": []}, "institutions": [{"id": "https://openalex.org/I14243506", "display_name": "Hong Kong Polytechnic University", "ror": "https://ror.org/0030zas98", "country_code": "HK", "type": "education", "lineage": ["https://openalex.org/I14243506"]}], "countries": ["HK"], "is_corresponding": false, "raw_author_name": "Xiao, Yaxin", "raw_affiliation_strings": ["The Hong Kong Polytechnic University , Hong Kong , China"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "The Hong Kong Polytechnic University , Hong Kong , China", "institution_ids": ["https://openalex.org/I14243506"]}]}, {"author_position": "last", "author": {"id": null, "display_name": "Li, Ronghua", "orcid": null, "observed_orcids": []}, "institutions": [{"id": "https://openalex.org/I14243506", "display_name": "Hong Kong Polytechnic University", "ror": "https://ror.org/0030zas98", "country_code": "HK", "type": "education", "lineage": ["https://openalex.org/I14243506"]}], "countries": ["HK"], "is_corresponding": false, "raw_author_name": "Li, Ronghua", "raw_affiliation_strings": ["The Hong Kong Polytechnic University , Hong Kong , China"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "The Hong Kong Polytechnic University , Hong Kong , China", "institution_ids": ["https://openalex.org/I14243506"]}]}], "abstract_inverted_index": {"Low": [0], "rank": [1, 108], "adaptation": [2], "(LoRA)": [3], "has": [4], "emerged": [5], "as": [6], "a": [7], "prominent": [8], "technique": [9], "for": [10], "fine-tuning": [11], "large": [12], "language": [13], "models": [14, 81], "(LLMs)": [15], "thanks": [16], "to": [17, 90, 102, 124, 134, 139], "its": [18, 37, 67, 111, 140], "superb": [19], "efficiency": [20], "gains": [21], "over": [22], "previous": [23], "methods.": [24], "While": [25], "extensive": [26], "studies": [27], "have": [28, 147], "examined": [29], "the": [30, 54, 64, 86, 92, 95], "performance": [31], "and": [32, 72, 98, 110], "structural": [33], "properties": [34], "of": [35, 57, 66, 94], "LoRA,": [36], "behavior": [38], "upon": [39], "training-time": [40, 114], "attacks": [41, 126], "remain": [42], "underexplored,": [43], "posing": [44], "significant": [45], "security": [46, 55], "risks.": [47], "In": [48], "this": [49], "paper,": [50], "we": [51], "theoretically": [52], "investigate": [53], "implications": [56], "LoRA's": [58, 82, 106], "low-rank": [59], "structure": [60, 109], "during": [61], "fine-tuning,": [62, 129], "in": [63], "context": [65], "robustness": [68, 123], "against": [69, 113], "data": [70, 136], "poisoning": [71, 137], "backdoor": [73, 125], "attacks.": [74, 115], "We": [75], "propose": [76], "an": [77], "analytical": [78], "framework": [79], "that": [80, 119], "training": [83, 96], "dynamics,": [84], "employs": [85], "neural": [87], "tangent": [88], "kernel": [89], "simplify": [91], "analysis": [93, 117], "process,": [97], "applies": [99], "information": [100, 142], "theory": [101], "establish": [103], "connections": [104], "between": [105], "low": [107], "vulnerability": [112], "Our": [116], "indicates": [118], "LoRA": [120], "exhibits": [121], "better": [122], "than": [127], "full": [128], "while": [130], "becomes": [131], "more": [132], "vulnerable": [133], "untargeted": [135], "due": [138], "over-simplified": [141], "geometry.": [143], "Extensive": [144], "experimental": [145], "evaluations": [146], "corroborated": [148], "our": [149], "theoretical": [150], "findings.": [151]}, "primary_topic": {"id": "https://openalex.org/T11689", "display_name": "Adversarial Robustness in Machine Learning", "score": 1.0, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T11689", "display_name": "Adversarial Robustness in Machine Learning", "score": 1.0, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10764", "display_name": "Privacy-Preserving Technologies in Data", "score": 0.0, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T11424", "display_name": "Security and Verification in Computing", "score": 0.0, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "pmh:oai:arXiv.org:2505.12871", "is_oa": true, "landing_page_url": "http://arxiv.org/abs/2505.12871", "pdf_url": "https://arxiv.org/pdf/2505.12871", "source": {"id": "https://openalex.org/S4306400194", "display_name": "arXiv (Cornell University)", "issn_l": "2331-8422", "issn": ["2331-8422"], "is_oa": true, "is_in_doaj": false, "is_core": false, "listed_in": [], "host_organization": "https://openalex.org/I205783295", "host_organization_name": "Cornell University", "host_organization_lineage": ["https://openalex.org/I205783295"], "host_organization_lineage_names": [], "type": "repository"}, "license": "cc-by", "license_id": "https://openalex.org/licenses/cc-by", "version": "submittedVersion", "is_accepted": false, "is_published": false, "raw_source_name": null, "raw_type": "text"}, "cited_by_count": 0, "is_retracted": false}, {"id": "https://openalex.org/W4407423028", "doi": "https://doi.org/10.48550/arxiv.2502.06846", "title": "Prot2Chat: Protein LLM with Early-Fusion of Text, Sequence and Structure", "publication_year": 2025, "publication_date": "2025-02-07", "type": "preprint", "language": "en", "authorships": [{"author_position": "first", "author": {"id": "https://openalex.org/A5005524411", "display_name": "Zhicong Wang", "orcid": "https://orcid.org/0000-0002-4238-9312", "observed_orcids": ["https://orcid.org/0000-0002-4238-9312"]}, "institutions": [{"id": "https://openalex.org/I3923682", "display_name": "Soochow University", "ror": "https://ror.org/05t8y2r12", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I3923682"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Wang, Zhicong", "raw_affiliation_strings": ["School of Computer Science and Technology , Soochow University , 215006 , Suzhou , China and"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "School of Computer Science and Technology , Soochow University , 215006 , Suzhou , China and", "institution_ids": ["https://openalex.org/I3923682"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5101191255", "display_name": "Zicheng Ma", "orcid": "https://orcid.org/0000-0002-2844-1217", "observed_orcids": ["https://orcid.org/0000-0002-2844-1217"]}, "institutions": [{"id": "https://openalex.org/I20231570", "display_name": "Peking University", "ror": "https://ror.org/02v51f717", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I20231570"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Ma, Zicheng", "raw_affiliation_strings": ["Changping Laboratory , Peking University , 102206 , Beijing , China"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Changping Laboratory , Peking University , 102206 , Beijing , China", "institution_ids": ["https://openalex.org/I20231570"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5009421684", "display_name": "Ziqiang Cao", "orcid": null, "observed_orcids": []}, "institutions": [{"id": "https://openalex.org/I3923682", "display_name": "Soochow University", "ror": "https://ror.org/05t8y2r12", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I3923682"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Cao, Ziqiang", "raw_affiliation_strings": ["School of Computer Science and Technology , Soochow University , 215006 , Suzhou , China and"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "School of Computer Science and Technology , Soochow University , 215006 , Suzhou , China and", "institution_ids": ["https://openalex.org/I3923682"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5023762288", "display_name": "Changlong Zhou", "orcid": "https://orcid.org/0000-0002-6387-4173", "observed_orcids": ["https://orcid.org/0000-0002-6387-4173", "https://orcid.org/0000-0001-6565-0379"]}, "institutions": [{"id": "https://openalex.org/I3923682", "display_name": "Soochow University", "ror": "https://ror.org/05t8y2r12", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I3923682"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Zhou, Changlong", "raw_affiliation_strings": ["School of Computer Science and Technology , Soochow University , 215006 , Suzhou , China and"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "School of Computer Science and Technology , Soochow University , 215006 , Suzhou , China and", "institution_ids": ["https://openalex.org/I3923682"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5100433265", "display_name": "Jun Zhang", "orcid": "https://orcid.org/0000-0003-2633-1521", "observed_orcids": ["https://orcid.org/0000-0003-2633-1521"]}, "institutions": [{"id": "https://openalex.org/I20231570", "display_name": "Peking University", "ror": "https://ror.org/02v51f717", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I20231570"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Zhang, Jun", "raw_affiliation_strings": ["Changping Laboratory , Peking University , 102206 , Beijing , China"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Changping Laboratory , Peking University , 102206 , Beijing , China", "institution_ids": ["https://openalex.org/I20231570"]}]}, {"author_position": "last", "author": {"id": null, "display_name": "Gao, Yiqin", "orcid": null, "observed_orcids": []}, "institutions": [{"id": "https://openalex.org/I20231570", "display_name": "Peking University", "ror": "https://ror.org/02v51f717", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I20231570"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Gao, Yiqin", "raw_affiliation_strings": ["Changping Laboratory , Peking University , 102206 , Beijing , China"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Changping Laboratory , Peking University , 102206 , Beijing , China", "institution_ids": ["https://openalex.org/I20231570"]}]}], "abstract_inverted_index": {"Motivation:": [0], "Proteins": [1], "are": [2, 168], "of": [3, 20, 26, 31, 37, 101, 153], "great": [4], "significance": [5], "in": [6, 65], "living": [7], "organisms.": [8], "However,": [9], "understanding": [10], "their": [11], "functions": [12], "encounters": [13], "numerous": [14], "challenges,": [15], "such": [16], "as": [17], "insufficient": [18], "integration": [19], "multimodal": [21], "information,": [22], "a": [23, 66, 71, 83], "large": [24, 72], "number": [25], "training": [27, 122], "parameters,": [28], "limited": [29], "flexibility": [30], "classification-based": [32], "methods,": [33], "and": [34, 62, 81, 103, 114, 128, 146, 156, 166], "the": [35, 51, 98, 107, 111, 115, 126, 135, 150], "lack": [36], "systematic": [38], "evaluation": [39], "metrics": [40, 145], "for": [41, 134], "protein": [42, 60, 88, 104], "Q&A": [43], "systems.": [44], "To": [45, 120], "tackle": [46], "these": [47, 95], "issues,": [48], "we": [49, 124], "propose": [50], "Prot2Chat": [52], "framework.": [53], "Results:": [54], "We": [55, 69], "modified": [56], "ProteinMPNN": [57], "to": [58, 76, 86, 117], "encode": [59, 77], "sequence": [61], "structural": [63], "information": [64, 89], "unified": [67], "way.": [68], "used": [70], "language": [73], "model": [74], "(LLM)": [75], "questions": [78, 116], "into": [79, 90], "vectors": [80], "developed": [82], "protein-text": [84], "adapter": [85], "compress": [87], "virtual": [91, 112], "tokens": [92, 113], "based": [93], "on": [94, 138], "vectors,": [96], "achieving": [97], "early": [99], "fusion": [100], "text": [102], "information.": [105], "Finally,": [106], "same": [108], "LLM": [109, 182], "reads": [110], "generate": [118], "answers.": [119], "optimize": [121], "efficiency,": [123], "froze": [125], "encoder": [127], "employed": [129], "Low-Rank": [130], "Adaptation": [131], "(LoRA)": [132], "techniques": [133], "LLM.": [136], "Experiments": [137], "two": [139], "datasets": [140], "show": [141], "that": [142], "both": [143], "automated": [144], "expert": [147], "evaluations": [148], "demonstrate": [149], "superior": [151], "performance": [152], "our": [154], "model,": [155], "zero-shot": [157], "prediction": [158], "results": [159], "highlight": [160], "its": [161], "generalization": [162], "ability.": [163], "The": [164], "models": [165], "codes": [167], "available": [169], "at": [170], "https://github.com/": [171], "wangzc1233/Prot2Chat.": [172], "Contact:": [173], "zqcao@suda.edu.cn": [174], "or": [175], "wangzc025@163.com": [176], "Key": [177], "words:": [178], "Protein": [179], "Q&A,": [180], "Early-Fusion,": [181]}, "primary_topic": {"id": "https://openalex.org/T12254", "display_name": "Machine Learning in Bioinformatics", "score": 0.8574000000953674, "subfield": {"id": "https://openalex.org/subfields/1312", "display_name": "Molecular Biology"}, "field": {"id": "https://openalex.org/fields/13", "display_name": "Biochemistry, Genetics and Molecular Biology"}, "domain": {"id": "https://openalex.org/domains/1", "display_name": "Life Sciences"}}, "topics": [{"id": "https://openalex.org/T12254", "display_name": "Machine Learning in Bioinformatics", "score": 0.8574000000953674, "subfield": {"id": "https://openalex.org/subfields/1312", "display_name": "Molecular Biology"}, "field": {"id": "https://openalex.org/fields/13", "display_name": "Biochemistry, Genetics and Molecular Biology"}, "domain": {"id": "https://openalex.org/domains/1", "display_name": "Life Sciences"}}, {"id": "https://openalex.org/T10044", "display_name": "Protein Structure and Dynamics", "score": 0.05480000004172325, "subfield": {"id": "https://openalex.org/subfields/1312", "display_name": "Molecular Biology"}, "field": {"id": "https://openalex.org/fields/13", "display_name": "Biochemistry, Genetics and Molecular Biology"}, "domain": {"id": "https://openalex.org/domains/1", "display_name": "Life Sciences"}}, {"id": "https://openalex.org/T11714", "display_name": "Multimodal Machine Learning Applications", "score": 0.03700000047683716, "subfield": {"id": "https://openalex.org/subfields/1707", "display_name": "Computer Vision and Pattern Recognition"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "pmh:oai:arXiv.org:2502.06846", "is_oa": true, "landing_page_url": "http://arxiv.org/abs/2502.06846", "pdf_url": "https://arxiv.org/pdf/2502.06846", "source": {"id": "https://openalex.org/S4306400194", "display_name": "arXiv (Cornell University)", "issn_l": "2331-8422", "issn": ["2331-8422"], "is_oa": true, "is_in_doaj": false, "is_core": false, "listed_in": [], "host_organization": "https://openalex.org/I205783295", "host_organization_name": "Cornell University", "host_organization_lineage": ["https://openalex.org/I205783295"], "host_organization_lineage_names": [], "type": "repository"}, "license": null, "license_id": null, "version": "submittedVersion", "is_accepted": false, "is_published": false, "raw_source_name": null, "raw_type": "text"}, "cited_by_count": 0, "is_retracted": false}, {"id": "https://openalex.org/W4413177717", "doi": "https://doi.org/10.30970/eli.30.3", "title": "PARAMETER EFFICIENT FINE-TUNING AND OVERFITTING IN GPT LARGE LANGUAGE MODELS: A METRIC-BASED COMPARISON", "publication_year": 2025, "publication_date": "2025-01-01", "type": "article", "language": "en", "authorships": [{"author_position": "first", "author": {"id": "https://openalex.org/A5026355556", "display_name": "Bohdan M. Pavlyshenko", "orcid": "https://orcid.org/0000-0001-9515-3488", "observed_orcids": ["https://orcid.org/0000-0001-9515-3488"]}, "institutions": [{"id": "https://openalex.org/I200280692", "display_name": "Lviv University", "ror": "https://ror.org/01s7y5e82", "country_code": "UA", "type": "education", "lineage": ["https://openalex.org/I200280692"]}], "countries": ["UA"], "is_corresponding": false, "raw_author_name": "Bohdan Pavlyshenko", "raw_affiliation_strings": ["Dr hab., Professor, System Design Department, Ivan Franko National University of Lviv, Ukaine"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Dr hab., Professor, System Design Department, Ivan Franko National University of Lviv, Ukaine", "institution_ids": ["https://openalex.org/I200280692"]}]}, {"author_position": "last", "author": {"id": "https://openalex.org/A5100284810", "display_name": "Ivan Bulka", "orcid": "https://orcid.org/0009-0003-2962-7931", "observed_orcids": ["https://orcid.org/0009-0003-2962-7931"]}, "institutions": [{"id": "https://openalex.org/I200280692", "display_name": "Lviv University", "ror": "https://ror.org/01s7y5e82", "country_code": "UA", "type": "education", "lineage": ["https://openalex.org/I200280692"]}], "countries": ["UA"], "is_corresponding": false, "raw_author_name": "Ivan Bulka", "raw_affiliation_strings": ["National University of Lviv , 50 Drahomanova St. , 79005 Lviv , Ukraine"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "National University of Lviv , 50 Drahomanova St. , 79005 Lviv , Ukraine", "institution_ids": ["https://openalex.org/I200280692"]}]}], "abstract_inverted_index": {"Background": [0], ".": [1, 113, 183, 257], "Building": [2], "upon": [3], "previous": [4], "research,": [5], "this": [6, 197, 275, 288], "study": [7, 69], "conducts": [8], "an": [9, 17, 65], "exploration": [10], "into": [11, 195, 273], "Large": [12], "Language": [13], "Models": [14], "(LLMs),": [15], "with": [16, 76, 200, 308], "emphasis": [18], "on": [19, 42, 46, 64, 131, 156], "the": [20, 51, 56, 59, 68, 72, 136, 167, 209, 221, 240, 243, 246, 249, 253, 265, 297, 303], "fine-tuning": [21, 115, 147, 188, 234, 281], "and": [22, 36, 53, 58, 86, 91, 111, 126, 174, 181, 189, 245, 278], "assessment": [23, 211], "of": [24, 62, 74, 103, 116, 139, 223, 242, 248, 261, 267, 287, 299, 305], "LLaMA-3.1": [25, 63, 85, 117, 191, 262], "for": [26, 177, 225, 282, 292], "instructional": [27], "tasks.": [28, 202, 217, 284], "LLaMA-3.1,": [29], "which": [30], "is": [31, 233], "a": [32, 82, 99, 235], "new": [33], "generation": [34], "model": [35, 198, 244, 276], "has": [37], "gained": [38], "considerable": [39], "recognition": [40], "based": [41], "its": [43, 88, 279], "superior": [44], "performance": [45, 214], "various": [47], "benchmarks.": [48], "Besides": [49], "assessing": [50, 212], "disparities": [52], "improvements": [54], "between": [55, 84], "base": [57], "fine-tuned": [60, 170], "versions": [61], "instruction": [66, 133, 216], "dataset,": [67], "also": [70], "addresses": [71], "concern": [73], "overfitting": [75, 254, 319], "LLaMA-3.1.": [77], "Furthermore,": [78], "it": [79], "carries": [80], "out": [81], "comparison": [83], "both": [87], "predecessor,": [89], "LLaMA-2,": [90], "another": [92], "LLM": [93, 140, 226], "known": [94], "as": [95, 122], "Mixtral,": [96, 316], "thereby": [97], "providing": [98], "more": [100, 293], "comprehensive": [101, 132], "picture": [102], "LLaMA-3.1's": [104], "capabilities": [105], "compared": [106], "to": [107, 239, 264], "other": [108], "models.": [109], "Materials": [110], "Methods": [112], "The": [114, 146, 184, 203, 218, 258, 285], "employed": [118], "state-of-the-art": [119], "techniques,": [120], "such": [121], "Low-Rank": [123, 128], "Adaptation": [124, 129], "(LoRA)": [125], "Quantized": [127], "(QLoRA),": [130], "datasets.": [134], "Acknowledging": [135], "resource-intensive": [137], "nature": [138, 241], "fine-tuning,": [141, 318], "optimization": [142], "measures": [143], "were": [144, 169], "taken.": [145], "process": [148], "was": [149], "additionally": [150], "enhanced": [151], "using": [152, 171], "Parameter-Efficient": [153], "Fine-tuning": [154], "(PEFT)": [155], "NVIDIA": [157], "A100": [158], "Tensor": [159], "Core": [160], "GPU": [161], "(graphics": [162], "processing": [163], "unit)": [164], "instances.": [165], "All": [166], "models": [168], "Hugging": [172], "Face": [173], "PyTorch": [175], "platforms": [176], "optimal": [178], "performance.": [179], "Results": [180], "Discussion": [182], "results": [185], "obtained": [186], "from": [187], "evaluating": [190], "offer": [192], "valuable": [193], "insights": [194, 272], "how": [196, 274], "performs": [199], "specific": [201], "evaluation": [204, 224, 307], "framework": [205], "proved": [206], "helpful": [207], "in": [208], "efficient": [210, 306], "LLMs'": [213], "concerning": [215], "research": [219, 289], "highlights": [220, 252, 302], "importance": [222, 304], "applications.": [227], "It": [228, 251, 301], "shows": [229], "that": [230], "not": [231], "always": [232], "good": [236], "choice,": [237], "due": [238], "specifics": [247], "task.": [250], "problem.": [255], "Conclusion": [256], "close": [259], "examination": [260], "contributes": [263], "field": [266], "machine": [268], "learning": [269], "by": [270], "offering": [271], "works": [277], "possible": [280], "special": [283], "findings": [286], "create": [290], "opportunities": [291], "in-depth": [294], "studies": [295], "around": [296], "application": [298], "LLMs.": [300], "already": [309], "designed": [310], "metrics.": [311], "Keywords": [312], ":": [313], "LLMs,": [314], "GPT,": [315], "LLaMA,": [317]}, "primary_topic": {"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.8246999979019165, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.8246999979019165, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T11307", "display_name": "Domain Adaptation and Few-Shot Learning", "score": 0.13040000200271606, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10181", "display_name": "Natural Language Processing Techniques", "score": 0.022299999371170998, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "doi:10.30970/eli.30.3", "is_oa": true, "landing_page_url": "https://doi.org/10.30970/eli.30.3", "pdf_url": "https://publications.lnu.edu.ua/collections/index.php/electronics/article/download/4887/5401", "source": {"id": "https://openalex.org/S4210185639", "display_name": "Electronics and Information Technologies", "issn_l": "2224-087X", "issn": ["2224-087X", "2224-0888"], "is_oa": true, "is_in_doaj": true, "is_core": false, "listed_in": ["doaj"], "host_organization": "https://openalex.org/P4383735918", "host_organization_name": "Lviv University", "host_organization_lineage": ["https://openalex.org/P4383735918"], "host_organization_lineage_names": ["Lviv University"], "type": "journal"}, "license": "cc-by", "license_id": "https://openalex.org/licenses/cc-by", "version": "publishedVersion", "is_accepted": true, "is_published": true, "raw_source_name": "Electronics and Information Technologies", "raw_type": "journal-article"}, "cited_by_count": 1, "is_retracted": false}, {"id": "https://openalex.org/W4415191234", "doi": "https://doi.org/10.48550/arxiv.2508.05731", "title": "InfiGUI-G1: Advancing GUI Grounding with Adaptive Exploration Policy Optimization", "publication_year": 2025, "publication_date": "2025-08-07", "type": "preprint", "language": "en", "authorships": [{"author_position": "first", "author": {"id": "https://openalex.org/A5100420950", "display_name": "Yuhang Liu", "orcid": "https://orcid.org/0000-0001-5687-184X", "observed_orcids": ["https://orcid.org/0000-0001-5687-184X"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Liu, Yuhang", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5052464770", "display_name": "Zeyu Liu", "orcid": "https://orcid.org/0000-0001-7318-9340", "observed_orcids": ["https://orcid.org/0000-0001-7318-9340"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Liu, Zeyu", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": null, "display_name": "Zhu, Shuanghe", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Zhu, Shuanghe", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5101469911", "display_name": "Pengxiang Li", "orcid": "https://orcid.org/0000-0003-4414-6186", "observed_orcids": ["https://orcid.org/0000-0003-4414-6186", "https://orcid.org/0009-0009-5595-442X", "https://orcid.org/0009-0009-2724-8771"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Li, Pengxiang", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": null, "display_name": "Xie, Congkai", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Xie, Congkai", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": null, "display_name": "Wang, Jiasheng", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Wang, Jiasheng", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": null, "display_name": "Hu, Xavier", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Hu, Xavier", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5101502419", "display_name": "Xiaotian Han", "orcid": "https://orcid.org/0000-0002-1344-3658", "observed_orcids": ["https://orcid.org/0000-0002-1344-3658"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Han, Xiaotian", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": null, "display_name": "Yuan, Jianbo", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Yuan, Jianbo", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5101412520", "display_name": "Xinyao Wang", "orcid": "https://orcid.org/0009-0008-8194-5669", "observed_orcids": ["https://orcid.org/0009-0008-8194-5669", "https://orcid.org/0000-0002-6159-7491"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Wang, Xinyao", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5100757084", "display_name": "Shengyu Zhang", "orcid": "https://orcid.org/0000-0001-6901-5874", "observed_orcids": ["https://orcid.org/0000-0001-6901-5874"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Zhang, Shengyu", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5101499561", "display_name": "Hongxia Yang", "orcid": "https://orcid.org/0000-0001-7461-4943", "observed_orcids": ["https://orcid.org/0000-0001-7461-4943"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Yang, Hongxia", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "last", "author": {"id": "https://openalex.org/A5004882141", "display_name": "Fei Qing Wu", "orcid": "https://orcid.org/0000-0003-2139-8807", "observed_orcids": ["https://orcid.org/0000-0003-2139-8807"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Wu, Fei", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}], "abstract_inverted_index": {"The": [0], "emergence": [1], "of": [2, 12, 46, 145, 168], "Multimodal": [3], "Large": [4], "Language": [5], "Models": [6], "(MLLMs)": [7], "has": [8, 73], "propelled": [9], "the": [10, 44, 58, 61, 173], "development": [11], "autonomous": [13], "agents": [14], "that": [15, 87], "operate": [16], "on": [17, 177], "Graphical": [18], "User": [19], "Interfaces": [20], "(GUIs)": [21], "using": [22], "pure": [23], "visual": [24], "input.": [25], "A": [26], "fundamental": [27], "challenge": [28], "is": [29, 129], "robustly": [30], "grounding": [31, 162], "natural": [32], "language": [33], "instructions.": [34], "This": [35], "requires": [36], "a": [37, 52, 113, 120, 133], "precise": [38], "spatial": [39, 80], "alignment,": [40, 55, 92], "which": [41, 56, 93, 128], "accurately": [42], "locates": [43], "coordinates": [45], "each": [47], "element,": [48], "and,": [49], "more": [50], "critically,": [51], "correct": [53], "semantic": [54, 91, 99, 184], "matches": [57], "instructions": [59], "to": [60, 75, 124, 170, 180], "functionally": [62], "appropriate": [63], "UI": [64], "element.": [65], "Although": [66], "Reinforcement": [67], "Learning": [68], "with": [69], "Verifiable": [70], "Rewards": [71], "(RLVR)": [72], "proven": [74], "be": [76], "effective": [77], "at": [78, 189], "improving": [79], "alignment": [81], "for": [82], "these": [83], "MLLMs,": [84], "we": [85, 106], "find": [86], "inefficient": [88], "exploration": [89, 104], "bottlenecks": [90], "prevent": [94], "models": [95], "from": [96, 142], "learning": [97], "difficult": [98], "associations.": [100], "To": [101], "address": [102], "this": [103], "problem,": [105], "present": [107], "Adaptive": [108, 136], "Exploration": [109, 137], "Policy": [110], "Optimization": [111], "(AEPO),": [112], "new": [114, 155], "policy": [115], "optimization": [116], "framework.": [117], "AEPO": [118], "employs": [119], "multi-answer": [121], "generation": [122], "strategy": [123], "enforce": [125], "broader": [126], "exploration,": [127], "then": [130], "guided": [131], "by": [132], "theoretically": [134], "grounded": [135], "Reward": [138], "(AER)": [139], "function": [140], "derived": [141], "first": [143], "principles": [144], "efficiency": [146], "eta=U/C.": [147], "Our": [148], "AEPO-trained": [149], "models,": [150], "InfiGUI-G1-3B": [151], "and": [152, 183], "InfiGUI-G1-7B,": [153], "establish": [154], "state-of-the-art": [156], "results": [157], "across": [158], "multiple": [159], "challenging": [160], "GUI": [161], "benchmarks,": [163], "achieving": [164], "significant": [165], "relative": [166], "improvements": [167], "up": [169], "9.0%": [171], "against": [172], "naive": [174], "RLVR": [175], "baseline": [176], "benchmarks": [178], "designed": [179], "test": [181], "generalization": [182], "understanding.": [185], "Resources": [186], "are": [187], "available": [188], "https://github.com/InfiXAI/InfiGUI-G1.": [190]}, "primary_topic": {"id": "https://openalex.org/T11714", "display_name": "Multimodal Machine Learning Applications", "score": 0.4884999990463257, "subfield": {"id": "https://openalex.org/subfields/1707", "display_name": "Computer Vision and Pattern Recognition"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T11714", "display_name": "Multimodal Machine Learning Applications", "score": 0.4884999990463257, "subfield": {"id": "https://openalex.org/subfields/1707", "display_name": "Computer Vision and Pattern Recognition"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10462", "display_name": "Reinforcement Learning in Robotics", "score": 0.08399999886751175, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T12031", "display_name": "Speech and dialogue systems", "score": 0.04399999976158142, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "pmh:oai:arXiv.org:2508.05731", "is_oa": true, "landing_page_url": "http://arxiv.org/abs/2508.05731", "pdf_url": "https://arxiv.org/pdf/2508.05731", "source": {"id": "https://openalex.org/S4306400194", "display_name": "arXiv (Cornell University)", "issn_l": "2331-8422", "issn": ["2331-8422"], "is_oa": true, "is_in_doaj": false, "is_core": false, "listed_in": [], "host_organization": "https://openalex.org/I205783295", "host_organization_name": "Cornell University", "host_organization_lineage": ["https://openalex.org/I205783295"], "host_organization_lineage_names": [], "type": "repository"}, "license": null, "license_id": null, "version": "submittedVersion", "is_accepted": false, "is_published": false, "raw_source_name": null, "raw_type": "text"}, "cited_by_count": 0, "is_retracted": false}, {"id": "https://openalex.org/W7115714448", "doi": "https://doi.org/10.2139/ssrn.5929216", "title": "ModiGen: A Large Language Model-Based Workflow for Multi-Domain System Modeling Language Generation", "publication_year": 2025, "publication_date": "2025-01-01", "type": "preprint", "language": null, "authorships": [{"author_position": "first", "author": {"id": null, "display_name": "Jiahui Xiang", "orcid": null, "observed_orcids": []}, "institutions": [{"id": "https://openalex.org/I76130692", "display_name": "Zhejiang University", "ror": "https://ror.org/00a2xv884", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I76130692"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Jiahui Xiang", "raw_affiliation_strings": ["Zhejiang University"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Zhejiang University", "institution_ids": ["https://openalex.org/I76130692"]}]}, {"author_position": "middle", "author": {"id": null, "display_name": "Tong Ye", "orcid": null, "observed_orcids": []}, "institutions": [{"id": "https://openalex.org/I76130692", "display_name": "Zhejiang University", "ror": "https://ror.org/00a2xv884", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I76130692"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Tong Ye", "raw_affiliation_strings": ["Zhejiang University"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Zhejiang University", "institution_ids": ["https://openalex.org/I76130692"]}]}, {"author_position": "middle", "author": {"id": null, "display_name": "Peiyu Liu", "orcid": null, "observed_orcids": []}, "institutions": [{"id": "https://openalex.org/I76130692", "display_name": "Zhejiang University", "ror": "https://ror.org/00a2xv884", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I76130692"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Peiyu Liu", "raw_affiliation_strings": ["Zhejiang University - College of Control Science and Engineering"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Zhejiang University - College of Control Science and Engineering", "institution_ids": ["https://openalex.org/I76130692"]}]}, {"author_position": "middle", "author": {"id": null, "display_name": "Yinan Zhang", "orcid": null, "observed_orcids": []}, "institutions": [{"id": "https://openalex.org/I76130692", "display_name": "Zhejiang University", "ror": "https://ror.org/00a2xv884", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I76130692"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Yinan Zhang", "raw_affiliation_strings": ["Zhejiang University"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Zhejiang University", "institution_ids": ["https://openalex.org/I76130692"]}]}, {"author_position": "last", "author": {"id": null, "display_name": "Wenhai Wang", "orcid": null, "observed_orcids": []}, "institutions": [{"id": "https://openalex.org/I76130692", "display_name": "Zhejiang University", "ror": "https://ror.org/00a2xv884", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I76130692"]}], "countries": ["CN"], "is_corresponding": true, "raw_author_name": "Wenhai Wang", "raw_affiliation_strings": ["Zhejiang University - College of Control Science and Engineering"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Zhejiang University - College of Control Science and Engineering", "institution_ids": ["https://openalex.org/I76130692"]}]}], "abstract_inverted_index": null, "primary_topic": {"id": "https://openalex.org/T13286", "display_name": "Modeling and Simulation Systems", "score": 0.9794999957084656, "subfield": {"id": "https://openalex.org/subfields/1703", "display_name": "Computational Theory and Mathematics"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T13286", "display_name": "Modeling and Simulation Systems", "score": 0.9794999957084656, "subfield": {"id": "https://openalex.org/subfields/1703", "display_name": "Computational Theory and Mathematics"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T11450", "display_name": "Model-Driven Software Engineering Techniques", "score": 0.011500000022351742, "subfield": {"id": "https://openalex.org/subfields/1712", "display_name": "Software"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.0027000000700354576, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "doi:10.2139/ssrn.5929216", "is_oa": true, "landing_page_url": "https://doi.org/10.2139/ssrn.5929216", "pdf_url": null, "source": {"id": "https://openalex.org/S4210172589", "display_name": "SSRN Electronic Journal", "issn_l": "1556-5068", "issn": ["1556-5068"], "is_oa": true, "is_in_doaj": false, "is_core": false, "listed_in": [], "host_organization": "https://openalex.org/I1318003438", "host_organization_name": "RELX Group (Netherlands)", "host_organization_lineage": ["https://openalex.org/I1318003438"], "host_organization_lineage_names": [], "type": "repository"}, "license": null, "license_id": null, "version": "acceptedVersion", "is_accepted": true, "is_published": false, "raw_source_name": null, "raw_type": "posted-content"}, "cited_by_count": 0, "is_retracted": false}, {"id": "https://openalex.org/W4410300964", "doi": "https://doi.org/10.1016/j.knosys.2025.113557", "title": "MoDeST: A dataset for Multi Domain Scientific Title Generation", "publication_year": 2025, "publication_date": "2025-05-12", "type": "article", "language": "en", "authorships": [{"author_position": "first", "author": {"id": "https://openalex.org/A5021857737", "display_name": "Necva B\u00f6l\u00fcc\u00fc", "orcid": "https://orcid.org/0000-0001-8121-3048", "observed_orcids": ["https://orcid.org/0000-0001-8121-3048"]}, "institutions": [{"id": "https://openalex.org/I1292875679", "display_name": "Commonwealth Scientific and Industrial Research Organisation", "ror": "https://ror.org/03qn8fb07", "country_code": "AU", "type": "government", "lineage": ["https://openalex.org/I1292875679", "https://openalex.org/I2801453606", "https://openalex.org/I4387156119"]}, {"id": "https://openalex.org/I42894916", "display_name": "Data61", "ror": "https://ror.org/03q397159", "country_code": "AU", "type": "other", "lineage": ["https://openalex.org/I1292875679", "https://openalex.org/I2801453606", "https://openalex.org/I42894916", "https://openalex.org/I4387156119"]}], "countries": ["AU"], "is_corresponding": true, "raw_author_name": "Necva B\u00f6l\u00fcc\u00fc", "raw_affiliation_strings": ["Data61, CSIRO, Sydney, 2122, NSW, Australia"], "raw_orcid": "https://orcid.org/0000-0001-8121-3048", "affiliations": [{"raw_affiliation_string": "Data61, CSIRO, Sydney, 2122, NSW, Australia", "institution_ids": ["https://openalex.org/I1292875679", "https://openalex.org/I42894916"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5033051572", "display_name": "Yunus Can Bilge", "orcid": "https://orcid.org/0000-0003-2811-3747", "observed_orcids": ["https://orcid.org/0000-0003-2811-3747"]}, "institutions": [{"id": "https://openalex.org/I66514158", "display_name": "Hacettepe University", "ror": "https://ror.org/04kwvgz42", "country_code": "TR", "type": "education", "lineage": ["https://openalex.org/I66514158"]}], "countries": ["TR"], "is_corresponding": false, "raw_author_name": "Yunus Can Bilge", "raw_affiliation_strings": ["Institute of Informatics, Hacettepe University, Ankara, 06800, Turkiye"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Institute of Informatics, Hacettepe University, Ankara, 06800, Turkiye", "institution_ids": ["https://openalex.org/I66514158"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5058278128", "display_name": "Dilber \u00c7etinta\u015f", "orcid": "https://orcid.org/0000-0003-0710-2280", "observed_orcids": ["https://orcid.org/0000-0003-0710-2280"]}, "institutions": [{"id": "https://openalex.org/I4210110665", "display_name": "Malatya Turgut \u00d6zal \u00dcniversitesi", "ror": "https://ror.org/01v2xem26", "country_code": "TR", "type": "education", "lineage": ["https://openalex.org/I4210110665"]}], "countries": ["TR"], "is_corresponding": false, "raw_author_name": "Dilber \u00c7etinta\u015f", "raw_affiliation_strings": ["University of Malatya Turgut \u00d6zal, Faculty of Engineering and Natural Sciences, Department of Computer Engineering, Malatya, 44210, Turkiye"], "raw_orcid": "https://orcid.org/0000-0003-0710-2280", "affiliations": [{"raw_affiliation_string": "University of Malatya Turgut \u00d6zal, Faculty of Engineering and Natural Sciences, Department of Computer Engineering, Malatya, 44210, Turkiye", "institution_ids": ["https://openalex.org/I4210110665"]}]}, {"author_position": "last", "author": {"id": "https://openalex.org/A5059457764", "display_name": "Zehra Y\u00fccel", "orcid": "https://orcid.org/0000-0002-2863-9119", "observed_orcids": ["https://orcid.org/0000-0002-2863-9119"]}, "institutions": [{"id": "https://openalex.org/I4210107659", "display_name": "Necmettin Erbakan University", "ror": "https://ror.org/013s3zh21", "country_code": "TR", "type": "education", "lineage": ["https://openalex.org/I4210107659"]}], "countries": ["TR"], "is_corresponding": false, "raw_author_name": "Zehra Y\u00fccel", "raw_affiliation_strings": ["Necmettin Erbakan University, Seydi\u015fehir Vocational School, Department of Computer Technologies, Konya, 42370, Turkiye"], "raw_orcid": "https://orcid.org/0000-0002-2863-9119", "affiliations": [{"raw_affiliation_string": "Necmettin Erbakan University, Seydi\u015fehir Vocational School, Department of Computer Technologies, Konya, 42370, Turkiye", "institution_ids": ["https://openalex.org/I4210107659"]}]}], "abstract_inverted_index": {"Title": [0], "generation": [1, 241], "is": [2, 192], "a": [3, 31, 64], "crucial": [4], "task": [5], "in": [6, 52, 118], "scientific": [7, 53], "writing,": [8], "serving": [9], "as": [10, 81], "the": [11, 93, 111, 140, 193, 207, 213, 245], "first": [12], "point": [13], "of": [14, 95, 113, 170, 209], "contact": [15], "for": [16, 145, 157, 176, 184, 198, 215, 218, 236], "readers": [17], "and": [18, 23, 34, 73, 87, 89, 98, 106, 109, 121, 130, 136, 166, 174, 178, 182, 206, 238, 249], "impacting": [19], "an": [20], "article\u2019s": [21], "visibility": [22], "citation": [24], "frequency.": [25], "In": [26], "this": [27], "paper,": [28], "we": [29, 188, 202], "present": [30], "novel": [32], "multi-domain": [33, 237], "multilingual": [35, 239], "dataset,": [36, 222], "MoDeST": [37, 69], "(": [38], "M": [39], "ulti-": [40], "D": [41], "omain": [42], "S": [43], "cientific": [44], "T": [45], "itle": [46], "Generation),": [47], "designed": [48], "to": [49, 63], "advance": [50], "research": [51], "title": [54, 147, 158, 240], "generation.": [55], "Unlike": [56], "existing": [57], "datasets,": [58], "which": [59], "are": [60], "often": [61], "limited": [62], "single": [65], "language": [66, 115], "or": [67], "domain,": [68], "includes": [70], "both": [71], "English": [72], "Turkish": [74], "titles": [75, 100], "across": [76, 229], "various": [77, 230], "academic": [78, 231], "disciplines,": [79, 232], "such": [80], "social": [82], "sciences,": [83], "medical": [84], "science,": [85], "science": [86], "engineering,": [88], "more.": [90], "We": [91, 126], "explore": [92], "challenges": [94, 205], "generating": [96, 199], "concise": [97], "informative": [99], "using": [101], "different": [102, 219], "sources": [103], "(keywords,": [104], "abstract,": [105], "full": [107], "article)": [108], "evaluate": [110], "performance": [112, 156], "large": [114], "models": [116, 217], "(LLMs)": [117], "zero-,": [119], "few-shot": [120], "supervised": [122], "fine-tuning": [123, 152], "(SFT)": [124], "settings.": [125], "conduct": [127], "human": [128, 134], "evaluation": [129, 143], "calculate": [131], "correlations": [132], "between": [133], "judgement": [135], "automatic": [137], "metrics,": [138], "identifying": [139], "most": [141, 194], "effective": [142, 195], "metric": [144], "assessing": [146], "quality.": [148], "Experiments": [149], "show": [150], "that": [151, 190], "significantly": [153], "improves": [154], "LLM": [155], "generation,": [159, 211], "with": [160, 223], "LLaMA-3.1": [161], "8B,": [162, 165], "70B,": [163], "Aya-expanse": [164], "32B": [167], "achieving": [168], "scores": [169], "40.12,": [171], "45.21,": [172], "45.31,": [173], "47.22": [175], "Turkish,": [177], "45.10,": [179], "49.10,": [180], "40.02,": [181], "48.54": [183], "English,": [185], "respectively.": [186], "Moreover,": [187], "find": [189], "abstract": [191], "input": [196], "source": [197], "titles.": [200], "Additionally,": [201], "analyse": [203], "domain-specific": [204], "impact": [208], "cross-lingual": [210], "highlighting": [212], "need": [214], "tailored": [216], "domains.": [220], "Our": [221], "its": [224, 234], "broad": [225], "representation,": [226], "ensures": [227], "applicability": [228], "enhancing": [233], "utility": [235], "while": [242], "also": [243], "benefiting": [244], "broader": [246], "NLP": [247], "community": [248], "related": [250], "tasks.": [251]}, "primary_topic": {"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.9589999914169312, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.9589999914169312, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T12648", "display_name": "Academic Writing and Publishing", "score": 0.014700000174343586, "subfield": {"id": "https://openalex.org/subfields/1207", "display_name": "History and Philosophy of Science"}, "field": {"id": "https://openalex.org/fields/12", "display_name": "Arts and Humanities"}, "domain": {"id": "https://openalex.org/domains/2", "display_name": "Social Sciences"}}, {"id": "https://openalex.org/T13083", "display_name": "Advanced Text Analysis Techniques", "score": 0.012400000356137753, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "doi:10.1016/j.knosys.2025.113557", "is_oa": true, "landing_page_url": "https://doi.org/10.1016/j.knosys.2025.113557", "pdf_url": null, "source": {"id": "https://openalex.org/S10169007", "display_name": "Knowledge-Based Systems", "issn_l": "0950-7051", "issn": ["0950-7051", "1872-7409"], "is_oa": false, "is_in_doaj": false, "is_core": true, "listed_in": ["cwts-core", "jufo-2", "norway-2"], "host_organization": "https://openalex.org/P4310320990", "host_organization_name": "Elsevier BV", "host_organization_lineage": ["https://openalex.org/P4310320990"], "host_organization_lineage_names": ["Elsevier BV"], "type": "journal"}, "license": "cc-by-nc", "license_id": "https://openalex.org/licenses/cc-by-nc", "version": "publishedVersion", "is_accepted": true, "is_published": true, "raw_source_name": "Knowledge-Based Systems", "raw_type": "journal-article"}, "cited_by_count": 1, "is_retracted": false}, {"id": "https://openalex.org/W4414756304", "doi": "https://doi.org/10.48550/arxiv.2509.06169", "title": "Reasoning Language Model for Personalized Lung Cancer Screening", "publication_year": 2025, "publication_date": "2025-09-07", "type": "preprint", "language": "en", "authorships": [{"author_position": "first", "author": {"id": "https://openalex.org/A5065193592", "display_name": "Chuang Niu", "orcid": "https://orcid.org/0000-0002-3310-7803", "observed_orcids": ["https://orcid.org/0000-0002-3310-7803"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Niu, Chuang", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "last", "author": {"id": "https://openalex.org/A5100400458", "display_name": "Ge Wang", "orcid": "https://orcid.org/0000-0002-2656-7705", "observed_orcids": ["https://orcid.org/0000-0002-2656-7705"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Wang, Ge", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}], "abstract_inverted_index": {"Accurate": [0], "risk": [1, 52, 62, 84, 108, 124, 134, 142], "assessment": [2], "in": [3, 107, 113], "lung": [4, 56, 82, 116, 170], "cancer": [5, 12, 83, 171], "screening": [6, 117], "is": [7], "critical": [8], "for": [9, 36, 80], "enabling": [10], "early": [11], "detection": [13], "and": [14, 24, 39, 47, 93, 99, 136, 156], "minimizing": [15], "unnecessary": [16], "invasive": [17], "procedures.": [18], "The": [19], "Lung": [20], "CT": [21], "Screening": [22], "Reporting": [23], "Data": [25], "System": [26], "(Lung-RADS)": [27], "has": [28], "been": [29], "widely": [30], "used": [31], "as": [32, 49], "the": [33, 114, 123, 130, 159], "standard": [34], "framework": [35], "patient": [37], "management": [38], "follow-up.": [40], "Nevertheless,": [41], "Lung-RADS": [42], "faces": [43], "trade-offs": [44], "between": [45], "sensitivity": [46], "specificity,": [48], "it": [50], "stratifies": [51], "solely": [53], "based": [54], "on": [55, 111], "nodule": [57], "characteristics": [58], "without": [59], "incorporating": [60], "various": [61], "factors.": [63], "Here": [64], "we": [65], "propose": [66], "a": [67, 87, 140], "reasoning": [68, 163], "language": [69], "model": [70, 103], "(RLM)": [71], "to": [72], "integrate": [73], "radiology": [74], "findings": [75], "with": [76], "longitudinal": [77], "medical": [78], "records": [79], "individualized": [81], "assessment.": [85], "Through": [86], "systematic": [88], "study": [89], "including": [90], "dataset": [91], "construction": [92], "distillation,": [94], "supervised": [95], "fine-tuning,": [96], "reinforcement": [97], "learning,": [98], "comprehensive": [100], "evaluation,": [101], "our": [102, 146], "makes": [104], "significant": [105], "improvements": [106], "prediction": [109], "performance": [110], "datasets": [112], "national": [115], "trial.": [118], "Notably,": [119], "RLM": [120], "can": [121], "decompose": [122], "evaluation": [125], "task": [126], "into": [127, 139, 169], "sub-components,": [128], "analyze": [129], "contributions": [131], "of": [132, 161], "diverse": [133], "factors,": [135], "synthesize": [137], "them": [138], "final": [141], "score": [143], "computed": [144], "using": [145], "data-driven": [147], "system": [148], "equation.": [149], "Our": [150], "approach": [151], "improves": [152], "both": [153], "predictive": [154], "accuracy": [155], "monitorability": [157], "through": [158], "chain": [160], "thought": [162], "process,": [164], "thereby": [165], "facilitating": [166], "clinical": [167], "translation": [168], "screening.": [172]}, "primary_topic": {"id": "https://openalex.org/T10202", "display_name": "Lung Cancer Diagnosis and Treatment", "score": 0.5565999746322632, "subfield": {"id": "https://openalex.org/subfields/2740", "display_name": "Pulmonary and Respiratory Medicine"}, "field": {"id": "https://openalex.org/fields/27", "display_name": "Medicine"}, "domain": {"id": "https://openalex.org/domains/4", "display_name": "Health Sciences"}}, "topics": [{"id": "https://openalex.org/T10202", "display_name": "Lung Cancer Diagnosis and Treatment", "score": 0.5565999746322632, "subfield": {"id": "https://openalex.org/subfields/2740", "display_name": "Pulmonary and Respiratory Medicine"}, "field": {"id": "https://openalex.org/fields/27", "display_name": "Medicine"}, "domain": {"id": "https://openalex.org/domains/4", "display_name": "Health Sciences"}}, {"id": "https://openalex.org/T10862", "display_name": "AI in cancer detection", "score": 0.29269999265670776, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T12422", "display_name": "Radiomics and Machine Learning in Medical Imaging", "score": 0.06830000132322311, "subfield": {"id": "https://openalex.org/subfields/2741", "display_name": "Radiology, Nuclear Medicine and Imaging"}, "field": {"id": "https://openalex.org/fields/27", "display_name": "Medicine"}, "domain": {"id": "https://openalex.org/domains/4", "display_name": "Health Sciences"}}], "primary_location": {"id": "pmh:oai:arXiv.org:2509.06169", "is_oa": true, "landing_page_url": "http://arxiv.org/abs/2509.06169", "pdf_url": "https://arxiv.org/pdf/2509.06169", "source": {"id": "https://openalex.org/S4306400194", "display_name": "arXiv (Cornell University)", "issn_l": "2331-8422", "issn": ["2331-8422"], "is_oa": true, "is_in_doaj": false, "is_core": false, "listed_in": [], "host_organization": "https://openalex.org/I205783295", "host_organization_name": "Cornell University", "host_organization_lineage": ["https://openalex.org/I205783295"], "host_organization_lineage_names": [], "type": "repository"}, "license": null, "license_id": null, "version": "submittedVersion", "is_accepted": false, "is_published": false, "raw_source_name": null, "raw_type": "text"}, "cited_by_count": 0, "is_retracted": false}, {"id": "https://openalex.org/W4416241498", "doi": "https://doi.org/10.48550/arxiv.2511.08567", "title": "The Path Not Taken: RLVR Provably Learns Off the Principals", "publication_year": 2025, "publication_date": "2025-11-11", "type": "preprint", "language": null, "authorships": [{"author_position": "first", "author": {"id": "https://openalex.org/A5031867835", "display_name": "Hanqing Zhu", "orcid": "https://orcid.org/0000-0002-2752-6753", "observed_orcids": ["https://orcid.org/0000-0002-2752-6753"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Zhu, Hanqing", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5100389531", "display_name": "Zhenyu Zhang", "orcid": "https://orcid.org/0000-0002-5200-3516", "observed_orcids": ["https://orcid.org/0000-0002-5200-3516", "https://orcid.org/0000-0003-0754-3676"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Zhang, Zhenyu", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5110397904", "display_name": "Hanxian Huang", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Huang, Hanxian", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5041890767", "display_name": "DiJia Su", "orcid": "https://orcid.org/0000-0001-9369-4715", "observed_orcids": ["https://orcid.org/0000-0001-9369-4715"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Su, DiJia", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5085241696", "display_name": "Zechun Liu", "orcid": "https://orcid.org/0000-0001-5856-2506", "observed_orcids": ["https://orcid.org/0000-0001-5856-2506"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Liu, Zechun", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": null, "display_name": "Zhao, Jiawei", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Zhao, Jiawei", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5056063665", "display_name": "Igor Fedorov", "orcid": "https://orcid.org/0000-0002-7235-4823", "observed_orcids": ["https://orcid.org/0000-0002-7235-4823"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Fedorov, Igor", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5031860916", "display_name": "Hamed Pirsiavash", "orcid": "https://orcid.org/0000-0002-2528-3305", "observed_orcids": ["https://orcid.org/0000-0002-2528-3305"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Pirsiavash, Hamed", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5110738653", "display_name": "Zhizhou Sha", "orcid": "https://orcid.org/0009-0004-2494-1945", "observed_orcids": ["https://orcid.org/0009-0004-2494-1945"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Sha, Zhizhou", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": null, "display_name": "Lee, Jinwon", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Lee, Jinwon", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5011883763", "display_name": "David Zhigang Pan", "orcid": "https://orcid.org/0000-0002-5705-2501", "observed_orcids": ["https://orcid.org/0000-0002-5705-2501"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Pan, David Z.", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": null, "display_name": "Wang, Zhangyang", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Wang, Zhangyang", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5106668475", "display_name": "Yuandong Tian", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Tian, Yuandong", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "last", "author": {"id": "https://openalex.org/A5034622870", "display_name": "Kai Sheng Tai", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Tai, Kai Sheng", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}], "abstract_inverted_index": {"Reinforcement": [0], "Learning": [1], "with": [2, 72], "Verifiable": [3], "Rewards": [4], "(RLVR)": [5], "reliably": [6], "improves": [7], "the": [8, 89, 108, 121, 160, 170, 236], "reasoning": [9], "performance": [10], "of": [11, 24, 38, 128, 174, 233, 238], "large": [12], "language": [13], "models,": [14], "yet": [15], "it": [16], "appears": [17], "to": [18, 51, 62], "modify": [19], "only": [20], "a": [21, 35, 39, 44, 73, 81, 125, 192, 227, 230], "small": [22], "fraction": [23], "parameters.": [25], "We": [26, 67, 114, 222], "revisit": [27], "this": [28, 117, 224], "paradox": [29], "and": [30, 59, 64, 98, 149, 162, 219, 235], "show": [31, 187], "that": [32, 188], "sparsity": [33], "is": [34], "surface": [36], "artifact": [37], "model-conditioned": [40], "optimization": [41, 194], "bias:": [42], "for": [43, 120], "fixed": [45], "pretrained": [46], "model,": [47], "updates": [48], "consistently": [49], "localize": [50], "preferred": [52], "parameter": [53], "regions,": [54, 106], "highly": [55], "consistent": [56], "across": [57], "runs": [58], "largely": [60], "invariant": [61], "datasets": [63], "RL": [65, 189], "recipes.": [66], "mechanistically": [68], "explain": [69], "these": [70, 167], "dynamics": [71], "Three-Gate": [74], "Theory:": [75], "Gate": [76, 84, 99], "I": [77], "(KL": [78], "Anchor)": [79], "imposes": [80], "KL-constrained": [82], "update;": [83], "II": [85], "(Model": [86], "Geometry)": [87], "steers": [88], "step": [90], "off": [91, 134], "principal": [92, 135, 157], "directions": [93, 136], "into": [94], "low-curvature,": [95], "spectrum-preserving": [96], "subspaces;": [97], "III": [100], "(Precision)": [101], "hides": [102], "micro-updates": [103], "in": [104, 137, 181, 191], "non-preferred": [105], "making": [107], "off-principal": [109, 150], "bias": [110], "appear": [111], "as": [112, 209], "sparsity.": [113], "then": [115], "validate": [116], "theory": [118], "and,": [119], "first": [122, 171], "time,": [123], "provide": [124, 169], "parameter-level": [126], "characterization": [127], "RLVR's": [129, 175], "learning": [130, 241], "dynamics:": [131], "RLVR": [132, 234], "learns": [133], "weight": [138], "space,": [139], "achieving": [140], "gains": [141], "via": [142], "minimal": [143], "spectral": [144], "drift,": [145], "reduced": [146], "principal-subspace": [147], "rotation,": [148], "update": [151], "alignment.": [152], "In": [153], "contrast,": [154], "SFT": [155], "targets": [156], "weights,": [158], "distorts": [159], "spectrum,": [161], "even": [163], "lags": [164], "RLVR.": [165], "Together,": [166], "results": [168], "parameter-space": [172], "account": [173], "training": [176], "dynamics,": [177], "revealing": [178], "clear": [179], "regularities": [180], "how": [182], "parameters": [183], "evolve.": [184], "Crucially,": [185], "we": [186], "operates": [190], "distinct": [193], "regime": [195], "from": [196], "SFT,": [197], "so": [198], "directly": [199], "adapting": [200], "SFT-era": [201, 246], "parameter-efficient": [202], "fine-tuning": [203, 218], "(PEFT)": [204], "methods": [205], "can": [206], "be": [207], "flawed,": [208], "evidenced": [210], "by": [211], "our": [212], "case": [213], "studies": [214], "on": [215], "advanced": [216], "sparse": [217], "LoRA": [220], "variants.": [221], "hope": [223], "work": [225], "charts": [226], "path": [228], "toward": [229], "white-box": [231], "understanding": [232], "design": [237], "geometry-aware,": [239], "RLVR-native": [240], "algorithms,": [242], "rather": [243], "than": [244], "repurposed": [245], "heuristics.": [247]}, "primary_topic": {"id": "https://openalex.org/T10462", "display_name": "Reinforcement Learning in Robotics", "score": 0.4332999885082245, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T10462", "display_name": "Reinforcement Learning in Robotics", "score": 0.4332999885082245, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T11307", "display_name": "Domain Adaptation and Few-Shot Learning", "score": 0.24379999935626984, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.1655000001192093, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "pmh:oai:arXiv.org:2511.08567", "is_oa": true, "landing_page_url": "http://arxiv.org/abs/2511.08567", "pdf_url": "https://arxiv.org/pdf/2511.08567", "source": {"id": "https://openalex.org/S4306400194", "display_name": "arXiv (Cornell University)", "issn_l": "2331-8422", "issn": ["2331-8422"], "is_oa": true, "is_in_doaj": false, "is_core": false, "listed_in": [], "host_organization": "https://openalex.org/I205783295", "host_organization_name": "Cornell University", "host_organization_lineage": ["https://openalex.org/I205783295"], "host_organization_lineage_names": [], "type": "repository"}, "license": null, "license_id": null, "version": "submittedVersion", "is_accepted": false, "is_published": false, "raw_source_name": null, "raw_type": "text"}, "cited_by_count": 0, "is_retracted": false}, {"id": "https://openalex.org/W4414902044", "doi": "https://doi.org/10.48550/arxiv.2503.14749", "title": "Uncertainty Distillation: Teaching Language Models to Express Semantic Confidence", "publication_year": 2025, "publication_date": "2025-03-18", "type": "preprint", "language": "en", "authorships": [{"author_position": "first", "author": {"id": "https://openalex.org/A5113264755", "display_name": "Sophia Hager", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Hager, Sophia", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5020643247", "display_name": "David M. Mueller", "orcid": "https://orcid.org/0000-0002-3851-2672", "observed_orcids": ["https://orcid.org/0000-0002-3851-2672"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Mueller, David", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5070418792", "display_name": "Kevin Duh", "orcid": "https://orcid.org/0000-0001-8107-4383", "observed_orcids": ["https://orcid.org/0000-0001-8107-4383"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Duh, Kevin", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "last", "author": {"id": "https://openalex.org/A5110927976", "display_name": "Nicholas Andrews", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Andrews, Nicholas", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}], "abstract_inverted_index": {"As": [0], "large": [1], "language": [2], "models": [3, 199], "(LLMs)": [4], "are": [5, 63, 179, 210], "increasingly": [6], "used": [7], "for": [8, 15, 72, 152], "factual": [9], "question-answering,": [10], "it": [11, 90, 96, 115], "becomes": [12], "more": [13, 99, 180, 212, 215], "important": [14], "LLMs": [16, 62], "to": [17, 21, 36, 54, 101, 125, 129, 137, 142, 173, 197], "have": [18], "the": [19, 23, 42, 46, 57, 70, 87, 106, 110], "capability": [20], "communicate": [22], "likelihood": [24], "that": [25, 157, 163, 191, 200, 209], "their": [26, 66], "answer": [27, 111], "is": [28, 116], "correct.": [29], "For": [30], "these": [31], "verbalized": [32, 150, 161], "expressions": [33], "of": [34, 49, 60, 113, 177, 207, 219], "uncertainty": [35, 73, 123, 140, 208], "be": [37, 98, 195], "meaningful,": [38], "they": [39], "should": [40], "reflect": [41], "error": [43, 58, 168], "rates": [44, 59], "at": [45, 185], "expressed": [47], "level": [48], "confidence.": [50], "However,": [51], "when": [52, 171], "prompted": [53], "express": [55], "confidence,": [56], "current": [61], "inconsistent": [64], "with": [65, 149, 166], "communicated": [67], "confidences,": [68], "highlighting": [69], "need": [71], "quantification": [74], "methods.": [75], "Many": [76], "prior": [77], "methods": [78], "calculate": [79], "lexical": [80], "uncertainty,": [81, 104], "estimating": [82], "a": [83, 120], "model's": [84, 107], "confidence": [85, 108], "in": [86, 109, 205], "specific": [88], "string": [89], "generated.": [91], "In": [92], "some": [93, 176], "cases,": [94], "however,": [95], "may": [97], "useful": [100], "estimate": [102], "semantic": [103, 132], "or": [105], "regardless": [112], "how": [114], "verbalized.": [117], "We": [118, 155], "propose": [119], "simple": [121], "procedure,": [122], "distillation,": [124], "teach": [126], "an": [127], "LLM": [128], "verbalize": [130], "calibrated": [131], "confidences.": [133], "Using": [134], "held-out": [135], "data": [136], "map": [138], "initial": [139], "estimates": [141, 206], "meaningful": [143], "probabilities,": [144], "we": [145, 189], "create": [146], "examples": [147], "annotated": [148], "probabilities": [151], "supervised": [153], "fine-tuning.": [154], "find": [156], "our": [158, 192, 220], "method": [159, 193], "yields": [160], "confidences": [162], "correlate": [164], "well": [165], "observed": [167], "rates,": [169], "even": [170], "compared": [172], "strong": [174], "baselines,": [175], "which": [178], "than": [181, 217], "twenty": [182], "times": [183], "slower": [184], "inference": [186], "time.": [187], "Additionally,": [188], "demonstrate": [190], "can": [194], "applied": [196], "black-box": [198], "allow": [201], "API-based": [202], "fine-tuning,": [203], "resulting": [204], "both": [211], "effective": [213], "and": [214], "efficient": [216], "any": [218], "baselines.": [221]}, "primary_topic": {"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.5220000147819519, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.5220000147819519, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T11689", "display_name": "Adversarial Robustness in Machine Learning", "score": 0.3280999958515167, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10181", "display_name": "Natural Language Processing Techniques", "score": 0.094200000166893, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "pmh:oai:arXiv.org:2503.14749", "is_oa": true, "landing_page_url": "http://arxiv.org/abs/2503.14749", "pdf_url": "https://arxiv.org/pdf/2503.14749", "source": {"id": "https://openalex.org/S4306400194", "display_name": "arXiv (Cornell University)", "issn_l": "2331-8422", "issn": ["2331-8422"], "is_oa": true, "is_in_doaj": false, "is_core": false, "listed_in": [], "host_organization": "https://openalex.org/I205783295", "host_organization_name": "Cornell University", "host_organization_lineage": ["https://openalex.org/I205783295"], "host_organization_lineage_names": [], "type": "repository"}, "license": null, "license_id": null, "version": "submittedVersion", "is_accepted": false, "is_published": false, "raw_source_name": null, "raw_type": "text"}, "cited_by_count": 0, "is_retracted": false}, {"id": "https://openalex.org/W4409152738", "doi": "https://doi.org/10.1038/s41592-025-02636-z", "title": "xTrimoPGLM: unified 100-billion-parameter pretrained transformer for deciphering the language of proteins", "publication_year": 2025, "publication_date": "2025-04-03", "type": "article", "language": "en", "authorships": [{"author_position": "first", "author": {"id": "https://openalex.org/A5101784459", "display_name": "Bo Chen", "orcid": "https://orcid.org/0000-0002-9629-5493", "observed_orcids": ["https://orcid.org/0000-0002-9629-5493", "https://orcid.org/0000-0003-4269-3063"]}, "institutions": [{"id": "https://openalex.org/I99065089", "display_name": "Tsinghua University", "ror": "https://ror.org/03cve4549", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I99065089"]}], "countries": ["CN"], "is_corresponding": true, "raw_author_name": "Bo Chen", "raw_affiliation_strings": ["BioMap Research, Palo Alto, CA, USA. cb21@mails.tsinghua.edu.cn", "Tsinghua University, Beijing, China. cb21@mails.tsinghua.edu.cn"], "raw_orcid": "https://orcid.org/0000-0002-9629-5493", "affiliations": [{"raw_affiliation_string": "BioMap Research, Palo Alto, CA, USA. cb21@mails.tsinghua.edu.cn", "institution_ids": []}, {"raw_affiliation_string": "Tsinghua University, Beijing, China. cb21@mails.tsinghua.edu.cn", "institution_ids": ["https://openalex.org/I99065089"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5074700282", "display_name": "Xingyi Cheng", "orcid": "https://orcid.org/0009-0005-4510-8693", "observed_orcids": ["https://orcid.org/0009-0005-4510-8693"]}, "institutions": [], "countries": [], "is_corresponding": true, "raw_author_name": "Xingyi Cheng", "raw_affiliation_strings": ["BioMap Research, Palo Alto, CA, USA. derrickzy@gmail.com"], "raw_orcid": "https://orcid.org/0009-0005-4510-8693", "affiliations": [{"raw_affiliation_string": "BioMap Research, Palo Alto, CA, USA. derrickzy@gmail.com", "institution_ids": []}]}, {"author_position": "middle", "author": {"id": null, "display_name": "Pan Li", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Pan Li", "raw_affiliation_strings": ["BioMap Research, Palo Alto, CA, USA"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "BioMap Research, Palo Alto, CA, USA", "institution_ids": []}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5018191388", "display_name": "Yangli\u2010ao Geng", "orcid": "https://orcid.org/0000-0003-0173-4041", "observed_orcids": ["https://orcid.org/0000-0003-0173-4041"]}, "institutions": [{"id": "https://openalex.org/I99065089", "display_name": "Tsinghua University", "ror": "https://ror.org/03cve4549", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I99065089"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Yangli-ao Geng", "raw_affiliation_strings": ["BioMap Research, Palo Alto, CA, USA", "Tsinghua University, Beijing, China"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "BioMap Research, Palo Alto, CA, USA", "institution_ids": []}, {"raw_affiliation_string": "Tsinghua University, Beijing, China", "institution_ids": ["https://openalex.org/I99065089"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5010443535", "display_name": "Jing Gong", "orcid": "https://orcid.org/0000-0002-9347-8008", "observed_orcids": ["https://orcid.org/0000-0002-9347-8008"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Jing Gong", "raw_affiliation_strings": ["BioMap Research, Palo Alto, CA, USA"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "BioMap Research, Palo Alto, CA, USA", "institution_ids": []}]}, {"author_position": "middle", "author": {"id": null, "display_name": "Shen Li", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Shen Li", "raw_affiliation_strings": ["BioMap Research, Palo Alto, CA, USA"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "BioMap Research, Palo Alto, CA, USA", "institution_ids": []}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5093688255", "display_name": "Zhilei Bei", "orcid": "https://orcid.org/0009-0008-7663-5551", "observed_orcids": ["https://orcid.org/0009-0008-7663-5551"]}, "institutions": [{"id": "https://openalex.org/I99065089", "display_name": "Tsinghua University", "ror": "https://ror.org/03cve4549", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I99065089"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Zhilei Bei", "raw_affiliation_strings": ["BioMap Research, Palo Alto, CA, USA", "Tsinghua University, Beijing, China"], "raw_orcid": "https://orcid.org/0009-0008-7663-5551", "affiliations": [{"raw_affiliation_string": "BioMap Research, Palo Alto, CA, USA", "institution_ids": []}, {"raw_affiliation_string": "Tsinghua University, Beijing, China", "institution_ids": ["https://openalex.org/I99065089"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5101323614", "display_name": "Xu Tan", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Xu Tan", "raw_affiliation_strings": ["BioMap Research, Palo Alto, CA, USA"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "BioMap Research, Palo Alto, CA, USA", "institution_ids": []}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5100736483", "display_name": "Boyan Wang", "orcid": "https://orcid.org/0000-0003-0088-2106", "observed_orcids": ["https://orcid.org/0000-0003-0088-2106"]}, "institutions": [{"id": "https://openalex.org/I99065089", "display_name": "Tsinghua University", "ror": "https://ror.org/03cve4549", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I99065089"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Boyan Wang", "raw_affiliation_strings": ["BioMap Research, Palo Alto, CA, USA", "Tsinghua University, Beijing, China"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "BioMap Research, Palo Alto, CA, USA", "institution_ids": []}, {"raw_affiliation_string": "Tsinghua University, Beijing, China", "institution_ids": ["https://openalex.org/I99065089"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5101553817", "display_name": "Xin Zeng", "orcid": "https://orcid.org/0000-0003-1014-2558", "observed_orcids": ["https://orcid.org/0000-0003-1014-2558"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Xin Zeng", "raw_affiliation_strings": ["BioMap Research, Palo Alto, CA, USA"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "BioMap Research, Palo Alto, CA, USA", "institution_ids": []}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5103106788", "display_name": "Chi-Ming Liu", "orcid": "https://orcid.org/0009-0003-3278-2983", "observed_orcids": ["https://orcid.org/0009-0003-3278-2983"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Chiming Liu", "raw_affiliation_strings": ["BioMap Research, Palo Alto, CA, USA"], "raw_orcid": "https://orcid.org/0009-0003-3278-2983", "affiliations": [{"raw_affiliation_string": "BioMap Research, Palo Alto, CA, USA", "institution_ids": []}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5005277155", "display_name": "Aohan Zeng", "orcid": "https://orcid.org/0000-0002-8766-0153", "observed_orcids": ["https://orcid.org/0000-0002-8766-0153"]}, "institutions": [{"id": "https://openalex.org/I99065089", "display_name": "Tsinghua University", "ror": "https://ror.org/03cve4549", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I99065089"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Aohan Zeng", "raw_affiliation_strings": ["Tsinghua University, Beijing, China"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Tsinghua University, Beijing, China", "institution_ids": ["https://openalex.org/I99065089"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5052284218", "display_name": "Yuxiao Dong", "orcid": "https://orcid.org/0000-0002-6092-2002", "observed_orcids": ["https://orcid.org/0000-0002-6092-2002"]}, "institutions": [{"id": "https://openalex.org/I99065089", "display_name": "Tsinghua University", "ror": "https://ror.org/03cve4549", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I99065089"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Yuxiao Dong", "raw_affiliation_strings": ["Tsinghua University, Beijing, China"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Tsinghua University, Beijing, China", "institution_ids": ["https://openalex.org/I99065089"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5044791875", "display_name": "Jie Tang", "orcid": "https://orcid.org/0000-0003-3487-4593", "observed_orcids": ["https://orcid.org/0000-0003-3487-4593"]}, "institutions": [{"id": "https://openalex.org/I99065089", "display_name": "Tsinghua University", "ror": "https://ror.org/03cve4549", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I99065089"]}], "countries": ["CN"], "is_corresponding": true, "raw_author_name": "Jie Tang", "raw_affiliation_strings": ["Tsinghua University, Beijing, China. jietang@tsinghua.edu.cn"], "raw_orcid": "https://orcid.org/0000-0003-3487-4593", "affiliations": [{"raw_affiliation_string": "Tsinghua University, Beijing, China. jietang@tsinghua.edu.cn", "institution_ids": ["https://openalex.org/I99065089"]}]}, {"author_position": "last", "author": {"id": "https://openalex.org/A5030589527", "display_name": "Le Song", "orcid": "https://orcid.org/0000-0002-9655-2787", "observed_orcids": ["https://orcid.org/0000-0002-9655-2787"]}, "institutions": [{"id": "https://openalex.org/I4210113480", "display_name": "Mohamed bin Zayed University of Artificial Intelligence", "ror": "https://ror.org/0258gkt32", "country_code": "AE", "type": "education", "lineage": ["https://openalex.org/I4210113480"]}], "countries": ["AE"], "is_corresponding": true, "raw_author_name": "Le Song", "raw_affiliation_strings": ["BioMap Research, Palo Alto, CA, USA. dasongle@gmail.com", "MBZUAI, Abu Dhabi, United Arab Emirates. dasongle@gmail.com"], "raw_orcid": "https://orcid.org/0000-0002-9655-2787", "affiliations": [{"raw_affiliation_string": "BioMap Research, Palo Alto, CA, USA. dasongle@gmail.com", "institution_ids": []}, {"raw_affiliation_string": "MBZUAI, Abu Dhabi, United Arab Emirates. dasongle@gmail.com", "institution_ids": ["https://openalex.org/I4210113480"]}]}], "abstract_inverted_index": null, "primary_topic": {"id": "https://openalex.org/T12254", "display_name": "Machine Learning in Bioinformatics", "score": 0.8345000147819519, "subfield": {"id": "https://openalex.org/subfields/1312", "display_name": "Molecular Biology"}, "field": {"id": "https://openalex.org/fields/13", "display_name": "Biochemistry, Genetics and Molecular Biology"}, "domain": {"id": "https://openalex.org/domains/1", "display_name": "Life Sciences"}}, "topics": [{"id": "https://openalex.org/T12254", "display_name": "Machine Learning in Bioinformatics", "score": 0.8345000147819519, "subfield": {"id": "https://openalex.org/subfields/1312", "display_name": "Molecular Biology"}, "field": {"id": "https://openalex.org/fields/13", "display_name": "Biochemistry, Genetics and Molecular Biology"}, "domain": {"id": "https://openalex.org/domains/1", "display_name": "Life Sciences"}}, {"id": "https://openalex.org/T10044", "display_name": "Protein Structure and Dynamics", "score": 0.1492999941110611, "subfield": {"id": "https://openalex.org/subfields/1312", "display_name": "Molecular Biology"}, "field": {"id": "https://openalex.org/fields/13", "display_name": "Biochemistry, Genetics and Molecular Biology"}, "domain": {"id": "https://openalex.org/domains/1", "display_name": "Life Sciences"}}, {"id": "https://openalex.org/T12576", "display_name": "vaccines and immunoinformatics approaches", "score": 0.0012000000569969416, "subfield": {"id": "https://openalex.org/subfields/1312", "display_name": "Molecular Biology"}, "field": {"id": "https://openalex.org/fields/13", "display_name": "Biochemistry, Genetics and Molecular Biology"}, "domain": {"id": "https://openalex.org/domains/1", "display_name": "Life Sciences"}}], "primary_location": {"id": "doi:10.1038/s41592-025-02636-z", "is_oa": false, "landing_page_url": "https://doi.org/10.1038/s41592-025-02636-z", "pdf_url": null, "source": {"id": "https://openalex.org/S127827428", "display_name": "Nature Methods", "issn_l": "1548-7091", "issn": ["1548-7091", "1548-7105"], "is_oa": false, "is_in_doaj": false, "is_core": true, "listed_in": ["cwts-core", "doyens", "jufo-3", "ki-jl-3", "medline", "norway-2"], "host_organization": "https://openalex.org/P4310319908", "host_organization_name": "Nature Portfolio", "host_organization_lineage": ["https://openalex.org/P4310319908", "https://openalex.org/P4310319965"], "host_organization_lineage_names": ["Nature Portfolio", "Springer Nature"], "type": "journal"}, "license": null, "license_id": null, "version": "publishedVersion", "is_accepted": true, "is_published": true, "raw_source_name": "Nature Methods", "raw_type": "journal-article"}, "cited_by_count": 56, "is_retracted": false}, {"id": "https://openalex.org/W4416210023", "doi": "https://doi.org/10.3390/electronics14224449", "title": "Modern Approaches to Software Vulnerability Detection: A Survey of Machine Learning, Deep Learning, and Large Language Models", "publication_year": 2025, "publication_date": "2025-11-14", "type": "article", "language": "en", "authorships": [{"author_position": "first", "author": {"id": "https://openalex.org/A5093214705", "display_name": "Md. Shazzad Hossain Shaon", "orcid": "https://orcid.org/0009-0008-2491-1622", "observed_orcids": ["https://orcid.org/0009-0008-2491-1622"]}, "institutions": [{"id": "https://openalex.org/I177721651", "display_name": "Oakland University", "ror": "https://ror.org/01ythxj32", "country_code": "US", "type": "education", "lineage": ["https://openalex.org/I177721651"]}], "countries": ["US"], "is_corresponding": false, "raw_author_name": "Md. Shazzad Hossain Shaon", "raw_affiliation_strings": ["Department of Computer Science and Engineering, Oakland University, Rochester, MI 48309, USA"], "raw_orcid": "https://orcid.org/0009-0008-2491-1622", "affiliations": [{"raw_affiliation_string": "Department of Computer Science and Engineering, Oakland University, Rochester, MI 48309, USA", "institution_ids": ["https://openalex.org/I177721651"]}]}, {"author_position": "last", "author": {"id": "https://openalex.org/A5015055011", "display_name": "Mst Shapna Akter", "orcid": "https://orcid.org/0000-0002-9859-6265", "observed_orcids": ["https://orcid.org/0000-0002-9859-6265"]}, "institutions": [{"id": "https://openalex.org/I177721651", "display_name": "Oakland University", "ror": "https://ror.org/01ythxj32", "country_code": "US", "type": "education", "lineage": ["https://openalex.org/I177721651"]}], "countries": ["US"], "is_corresponding": true, "raw_author_name": "Mst Shapna Akter", "raw_affiliation_strings": ["Department of Computer Science and Engineering, Oakland University, Rochester, MI 48309, USA"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Department of Computer Science and Engineering, Oakland University, Rochester, MI 48309, USA", "institution_ids": ["https://openalex.org/I177721651"]}]}], "abstract_inverted_index": {"Software": [0], "vulnerabilities": [1], "pose": [2], "significant": [3], "risks": [4], "to": [5, 80, 162, 199], "the": [6, 33, 106, 154, 184], "security": [7], "and": [8, 23, 30, 53, 73, 84, 97, 120, 145, 174, 191, 203], "reliability": [9], "of": [10, 35, 46, 87, 102, 108], "modern": [11], "systems,": [12], "making": [13], "automated": [14], "vulnerability": [15, 60, 113, 149, 206], "detection": [16, 207], "an": [17], "essential": [18], "research": [19, 132], "area.": [20], "Traditional": [21], "static": [22], "rule-based": [24], "approaches": [25, 190], "are": [26], "limited": [27, 112], "in": [28, 66], "scalability": [29], "adaptability,": [31], "motivating": [32], "adoption": [34], "data-driven": [36], "methods.": [37], "In": [38], "this": [39], "survey,": [40], "we": [41, 91, 128], "present": [42, 155], "a": [43, 99], "comprehensive": [44, 165], "review": [45], "Machine": [47], "Learning": [48, 51], "(ML),": [49], "Deep": [50], "(DL),": [52], "Large": [54], "Language": [55], "Models": [56], "(LLMs)": [57], "techniques": [58], "for": [59, 148], "detection.": [61, 150], "We": [62], "analyze": [63], "recent": [64], "advances": [65], "feature": [67, 172], "representation,": [68], "fine-tuning": [69], "strategies,": [70], "generative": [71], "approaches,": [72], "prompt": [74], "engineering,": [75], "while": [76], "highlighting": [77], "their": [78], "ability": [79], "capture": [81], "both": [82], "syntactic": [83], "semantic": [85], "properties": [86], "source": [88], "code.": [89], "Furthermore,": [90], "examine": [92], "commonly": [93], "used": [94], "evaluation": [95, 166], "metrics": [96], "provide": [98], "critical": [100], "discussion": [101], "key": [103], "challenges,": [104], "including": [105], "lack": [107], "large-scale": [109], "real-world": [110], "datasets,": [111], "coverage,": [114], "class": [115], "imbalance,": [116], "interpretability": [117], "gaps,": [118], "hallucination,": [119], "high": [121], "computational": [122], "costs.": [123], "To": [124], "address": [125], "these": [126, 179], "issues,": [127], "outline": [129], "promising": [130], "future": [131], "directions,": [133], "such": [134], "as": [135], "neuro-symbolic": [136], "hybrid": [137], "methods,": [138], "parameter-efficient": [139], "fine-tuning,": [140], "continual": [141], "learning,": [142], "cross-language": [143], "generalization,": [144], "explainable": [146], "AI": [147], "Unlike": [151], "previous": [152], "studies,": [153], "work": [156], "explores": [157], "learning": [158], "paradigms": [159], "from": [160], "ML": [161], "LLMs": [163], "using": [164], "criteria": [167], "that": [168], "highlight": [169], "analytical": [170], "capability,": [171], "interpretability,": [173], "code-context": [175], "comprehension.": [176], "By": [177], "combining": [178], "factors,": [180], "our": [181], "study": [182], "addresses": [183], "methodological": [185], "gap": [186], "between": [187], "classic": [188], "feature-based": [189], "current": [192], "LLM-driven": [193], "reasoning": [194], "frameworks,": [195], "providing": [196], "beneficial": [197], "insights": [198], "develop": [200], "robust,": [201], "scalable,": [202], "trustworthy": [204], "software": [205], "systems.": [208]}, "primary_topic": {"id": "https://openalex.org/T10260", "display_name": "Software Engineering Research", "score": 0.5278000235557556, "subfield": {"id": "https://openalex.org/subfields/1710", "display_name": "Information Systems"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T10260", "display_name": "Software Engineering Research", "score": 0.5278000235557556, "subfield": {"id": "https://openalex.org/subfields/1710", "display_name": "Information Systems"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10743", "display_name": "Software Testing and Debugging Techniques", "score": 0.32280001044273376, "subfield": {"id": "https://openalex.org/subfields/1712", "display_name": "Software"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T12479", "display_name": "Web Application Security Vulnerabilities", "score": 0.09179999679327011, "subfield": {"id": "https://openalex.org/subfields/1710", "display_name": "Information Systems"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "doi:10.3390/electronics14224449", "is_oa": true, "landing_page_url": "https://doi.org/10.3390/electronics14224449", "pdf_url": "https://www.mdpi.com/2079-9292/14/22/4449/pdf?version=1763126210", "source": {"id": "https://openalex.org/S4210202905", "display_name": "Electronics", "issn_l": "2079-9292", "issn": ["2079-9292"], "is_oa": true, "is_in_doaj": false, "is_core": true, "listed_in": ["cwts-core", "norway-1"], "host_organization": "https://openalex.org/P4310310987", "host_organization_name": "Multidisciplinary Digital Publishing Institute", "host_organization_lineage": ["https://openalex.org/P4310310987"], "host_organization_lineage_names": ["Multidisciplinary Digital Publishing Institute"], "type": "journal"}, "license": "cc-by", "license_id": "https://openalex.org/licenses/cc-by", "version": "publishedVersion", "is_accepted": true, "is_published": true, "raw_source_name": "Electronics", "raw_type": "journal-article"}, "cited_by_count": 9, "is_retracted": false}, {"id": "https://openalex.org/W4415937532", "doi": "https://doi.org/10.48550/arxiv.2511.00405", "title": "UME-R1: Exploring Reasoning-Driven Generative Multimodal Embeddings", "publication_year": 2025, "publication_date": "2025-11-01", "type": "preprint", "language": null, "authorships": [{"author_position": "first", "author": {"id": "https://openalex.org/A5112974306", "display_name": "Zhibin Lan", "orcid": "https://orcid.org/0009-0008-3930-3101", "observed_orcids": ["https://orcid.org/0009-0008-3930-3101"]}, "institutions": [{"id": "https://openalex.org/I191208505", "display_name": "Xiamen University", "ror": "https://ror.org/00mcjh785", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I191208505"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Lan, Zhibin", "raw_affiliation_strings": ["School of Informatics , Xiamen University , China,"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "School of Informatics , Xiamen University , China,", "institution_ids": ["https://openalex.org/I191208505"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5005680009", "display_name": "Liqiang Niu", "orcid": "https://orcid.org/0000-0002-3366-9486", "observed_orcids": ["https://orcid.org/0000-0002-3366-9486"]}, "institutions": [{"id": "https://openalex.org/I2250653659", "display_name": "Tencent (China)", "ror": "https://ror.org/00hhjss72", "country_code": "CN", "type": "company", "lineage": ["https://openalex.org/I2250653659"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Niu, Liqiang", "raw_affiliation_strings": ["WeChat AI , Tencent Inc , China,"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "WeChat AI , Tencent Inc , China,", "institution_ids": ["https://openalex.org/I2250653659"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5024849044", "display_name": "Fandong Meng", "orcid": "https://orcid.org/0000-0002-8158-2377", "observed_orcids": ["https://orcid.org/0000-0002-8158-2377"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Meng, Fandong", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5055646953", "display_name": "Jie Na Zhou", "orcid": "https://orcid.org/0000-0002-6002-291X", "observed_orcids": ["https://orcid.org/0000-0002-6002-291X", "https://orcid.org/0009-0005-4708-2730"]}, "institutions": [{"id": "https://openalex.org/I2250653659", "display_name": "Tencent (China)", "ror": "https://ror.org/00hhjss72", "country_code": "CN", "type": "company", "lineage": ["https://openalex.org/I2250653659"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Zhou, Jie", "raw_affiliation_strings": ["WeChat AI , Tencent Inc , China,"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "WeChat AI , Tencent Inc , China,", "institution_ids": ["https://openalex.org/I2250653659"]}]}, {"author_position": "last", "author": {"id": null, "display_name": "Su, Jinsong", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Su, Jinsong", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}], "abstract_inverted_index": {"The": [0], "remarkable": [1], "success": [2], "of": [3, 37, 56, 118, 134, 164], "multimodal": [4, 13, 52, 197], "large": [5], "language": [6], "models": [7, 17, 187], "(MLLMs)": [8], "has": [9], "driven": [10], "advances": [11], "in": [12], "embeddings,": [14, 39, 143], "yet": [15], "existing": [16], "remain": [18], "inherently": [19], "discriminative,": [20], "limiting": [21], "their": [22], "ability": [23], "to": [24, 74], "benefit": [25], "from": [26], "reasoning-driven": [27, 195], "generation": [28], "paradigm.": [29, 46], "In": [30], "this": [31], "work,": [32], "we": [33], "pioneer": [34], "the": [35, 66, 113, 160, 169], "exploration": [36], "generative": [38, 45, 79, 90, 101, 115, 123, 142, 165, 196], "unifying": [40], "embedding": [41, 53, 91, 186], "tasks": [42, 174], "within": [43], "a": [44, 50, 57, 61, 81, 145, 190], "We": [47], "propose": [48], "UME-R1,": [49], "universal": [51], "framework": [54], "consisting": [55], "two-stage": [58], "training": [59], "strategy:": [60], "cold-start": [62], "supervised": [63], "fine-tuning": [64], "equips": [65], "model": [67], "with": [68], "reasoning": [69, 86, 116], "capabilities": [70, 117], "and": [71, 78, 87, 122, 178, 188, 202], "enables": [72], "it": [73], "generate": [75], "both": [76], "discriminative": [77, 109, 121, 185], "embeddings;": [80], "subsequent": [82], "reinforcement": [83], "learning": [84], "enhances": [85], "further": [88], "optimizes": [89], "quality.": [92], "This": [93], "pioneering": [94], "work": [95], "reveals": [96], "four": [97], "key": [98], "insights:": [99], "1)": [100], "embeddings": [102, 110, 124], "unlock": [103], "substantial": [104], "performance": [105, 130], "gains": [106], "over": [107], "conventional": [108, 184], "by": [111], "leveraging": [112], "powerful": [114], "MLLMs;": [119], "2)": [120], "are": [125], "complementary,": [126], "whose": [127], "combined": [128], "oracle": [129], "far": [131], "exceeding": [132], "that": [133], "either": [135], "alone;": [136], "3)": [137], "RL": [138], "can": [139], "effectively": [140], "enhance": [141], "establishing": [144], "scalable": [146], "optimization": [147], "paradigm.;": [148], "4)": [149], "repeated": [150], "sampling": [151], "at": [152, 208], "inference": [153], "boosts": [154], "downstream": [155], "task": [156], "coverage": [157], "(pass@k),": [158], "highlighting": [159], "inference-time": [161], "scalability": [162], "potential": [163], "embeddings.": [166, 198], "Evaluated": [167], "on": [168], "MMEB-V2": [170], "benchmark": [171], "across": [172], "78": [173], "spanning": [175], "video,": [176], "image,": [177], "visual": [179], "documents,": [180], "UME-R1": [181], "significantly": [182], "outperforms": [183], "offers": [189], "foundation": [191], "for": [192], "more": [193], "interpretable,": [194], "Our": [199], "code,": [200], "models,": [201], "datasets": [203], "will": [204], "be": [205], "publicly": [206], "available": [207], "https://github.com/XMUDeepLIT/UME-R1.": [209]}, "primary_topic": {"id": "https://openalex.org/T11714", "display_name": "Multimodal Machine Learning Applications", "score": 1.0, "subfield": {"id": "https://openalex.org/subfields/1707", "display_name": "Computer Vision and Pattern Recognition"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T11714", "display_name": "Multimodal Machine Learning Applications", "score": 1.0, "subfield": {"id": "https://openalex.org/subfields/1707", "display_name": "Computer Vision and Pattern Recognition"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.0, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10627", "display_name": "Advanced Image and Video Retrieval Techniques", "score": 0.0, "subfield": {"id": "https://openalex.org/subfields/1707", "display_name": "Computer Vision and Pattern Recognition"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "pmh:oai:arXiv.org:2511.00405", "is_oa": true, "landing_page_url": "http://arxiv.org/abs/2511.00405", "pdf_url": "https://arxiv.org/pdf/2511.00405", "source": {"id": "https://openalex.org/S4306400194", "display_name": "arXiv (Cornell University)", "issn_l": "2331-8422", "issn": ["2331-8422"], "is_oa": true, "is_in_doaj": false, "is_core": false, "listed_in": [], "host_organization": "https://openalex.org/I205783295", "host_organization_name": "Cornell University", "host_organization_lineage": ["https://openalex.org/I205783295"], "host_organization_lineage_names": [], "type": "repository"}, "license": null, "license_id": null, "version": "submittedVersion", "is_accepted": false, "is_published": false, "raw_source_name": null, "raw_type": "text"}, "cited_by_count": 0, "is_retracted": false}, {"id": "https://openalex.org/W4408990501", "doi": "https://doi.org/10.54097/we271q84", "title": "Efficient Differentially Private Fine-Tuning with QLoRA and Prefix Tuning for Large Language Models", "publication_year": 2025, "publication_date": "2025-03-27", "type": "article", "language": "en", "authorships": [{"author_position": "first", "author": {"id": null, "display_name": "Zhouyi Tan", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Zhouyi Tan", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5135180768", "display_name": "Xi Xiong", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Xi Xiong", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "last", "author": {"id": "https://openalex.org/A5101515722", "display_name": "Dong Xu", "orcid": "https://orcid.org/0000-0002-7085-7726", "observed_orcids": ["https://orcid.org/0000-0002-7085-7726", "https://orcid.org/0009-0005-5641-2535"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Dong Xu", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}], "abstract_inverted_index": {"Large": [0], "language": [1, 10], "models": [2, 25], "(LLMs)": [3], "have": [4], "achieved": [5], "remarkable": [6], "success": [7], "in": [8, 218], "natural": [9], "processing": [11], "(NLP)": [12], "tasks.": [13], "However,": [14], "fine-tuning": [15, 71, 204, 217], "LLMs": [16], "using": [17, 155, 168], "private": [18, 70], "datasets": [19, 157], "raises": [20], "significant": [21], "privacy": [22, 52, 144, 198], "concerns,": [23], "as": [24], "can": [26], "inadvertently": [27], "memorize": [28], "sensitive": [29], "information.": [30], "Differentially": [31], "Private": [32], "Stochastic": [33], "Gradient": [34], "Descent": [35], "(DP-SGD)": [36], "provides": [37, 208], "a": [38, 101, 123, 196, 209], "mathematically": [39], "rigorous": [40], "solution": [41, 213], "but": [42], "suffers": [43], "from": [44], "high": [45], "computational": [46, 96], "overhead,": [47], "slow": [48], "convergence,": [49], "and": [50, 79, 87, 94, 137, 146, 164, 171, 211], "excessive": [51], "budget": [53, 199], "consumption,": [54], "making": [55], "it": [56], "impractical": [57], "for": [58, 134, 140, 214], "large-scale": [59], "models.": [60], "To": [61], "address": [62], "these": [63], "challenges,": [64], "we": [65, 121], "propose": [66], "an": [67], "efficient": [68], "differentially": [69], "method": [72], "that": [73, 175], "combines": [74], "Quantized": [75], "Low-Rank": [76], "Adaptation": [77], "(QLoRA)": [78], "Prefix": [80, 98, 135, 178], "Tuning.": [81], "QLoRA": [82, 176], "employs": [83], "4-bit": [84], "NormalFloat": [85], "quantization": [86], "low-rank": [88], "adaptation,": [89], "significantly": [90], "reducing": [91, 114], "memory": [92, 184], "consumption": [93], "improving": [95], "efficiency.": [97], "Tuning": [99, 136, 179], "optimizes": [100], "small": [102], "set": [103], "of": [104, 117, 190], "prefix": [105], "vectors": [106], "without": [107], "modifying": [108], "the": [109, 115, 191], "model\u2019s": [110], "main": [111], "parameters,": [112], "further": [113], "impact": [116], "DP": [118, 203], "noise.": [119], "Additionally,": [120], "introduce": [122], "hybrid": [124], "adaptive": [125, 132], "gradient": [126], "clipping": [127, 133, 139], "strategy,": [128], "which": [129], "applies": [130], "sample-wise": [131], "group-wise": [138], "QLoRA,": [141], "effectively": [142], "balancing": [143], "protection": [145], "model": [147, 193], "utility.": [148], "We": [149], "evaluate": [150], "our": [151], "approach": [152], "on": [153], "GPT-2": [154], "benchmark": [156], "including": [158], "E2E": [159], "NLG": [160], "Challenge,": [161], "XSum,": [162], "SST-2,": [163], "DART,": [165], "measuring": [166], "performance": [167, 194], "BLEU,": [169], "ROUGE,": [170], "F1-score.": [172], "Results": [173], "demonstrate": [174], "+": [177], "achieves": [180], "up": [181], "to": [182], "75%": [183], "reduction": [185], "while": [186], "maintaining": [187], "over": [188], "95%": [189], "original": [192], "under": [195], "moderate": [197], "(\u03b5=3),": [200], "outperforming": [201], "traditional": [202], "methods.": [205], "Our": [206], "work": [207], "practical": [210], "scalable": [212], "privacy-preserving": [215], "LLM": [216], "resource-constrained": [219], "environments.": [220]}, "primary_topic": {"id": "https://openalex.org/T10764", "display_name": "Privacy-Preserving Technologies in Data", "score": 0.9980000257492065, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T10764", "display_name": "Privacy-Preserving Technologies in Data", "score": 0.9980000257492065, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.00039999998989515007, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T11689", "display_name": "Adversarial Robustness in Machine Learning", "score": 0.00019999999494757503, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "doi:10.54097/we271q84", "is_oa": true, "landing_page_url": "https://doi.org/10.54097/we271q84", "pdf_url": null, "source": {"id": "https://openalex.org/S5407049135", "display_name": "Journal of Computer Science and Artificial Intelligence", "issn_l": "3078-2309", "issn": ["3078-2309", "3078-8242"], "is_oa": false, "is_in_doaj": false, "is_core": false, "listed_in": [], "host_organization": null, "host_organization_name": null, "host_organization_lineage": [], "host_organization_lineage_names": [], "type": "journal"}, "license": "cc-by-nc", "license_id": "https://openalex.org/licenses/cc-by-nc", "version": "publishedVersion", "is_accepted": true, "is_published": true, "raw_source_name": "Journal of Computer Science and Artificial Intelligence", "raw_type": "journal-article"}, "cited_by_count": 0, "is_retracted": false}, {"id": "https://openalex.org/W4416365886", "doi": "https://doi.org/10.48550/arxiv.2508.00410", "title": "Co-rewarding: Stable Self-supervised RL for Eliciting Reasoning in Large Language Models", "publication_year": 2025, "publication_date": "2025-08-01", "type": "preprint", "language": "en", "authorships": [{"author_position": "first", "author": {"id": "https://openalex.org/A5067204895", "display_name": "Zizhuo Zhang", "orcid": "https://orcid.org/0000-0003-1943-5963", "observed_orcids": ["https://orcid.org/0000-0003-1943-5963"]}, "institutions": [{"id": "https://openalex.org/I141568987", "display_name": "Hong Kong Baptist University", "ror": "https://ror.org/0145fw131", "country_code": "HK", "type": "education", "lineage": ["https://openalex.org/I141568987"]}, {"id": "https://openalex.org/I80804394", "display_name": "Southwest Baptist University", "ror": "https://ror.org/02dgbcj80", "country_code": "US", "type": "education", "lineage": ["https://openalex.org/I80804394"]}], "countries": ["HK", "US"], "is_corresponding": false, "raw_author_name": "Zhang, Zizhuo", "raw_affiliation_strings": ["TMLR Group , Department of Computer Science , Hong Kong Baptist University"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "TMLR Group , Department of Computer Science , Hong Kong Baptist University", "institution_ids": ["https://openalex.org/I141568987", "https://openalex.org/I80804394"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5106550627", "display_name": "Jianing Zhu", "orcid": "https://orcid.org/0000-0002-0100-5562", "observed_orcids": ["https://orcid.org/0000-0002-0100-5562"]}, "institutions": [{"id": "https://openalex.org/I141568987", "display_name": "Hong Kong Baptist University", "ror": "https://ror.org/0145fw131", "country_code": "HK", "type": "education", "lineage": ["https://openalex.org/I141568987"]}, {"id": "https://openalex.org/I80804394", "display_name": "Southwest Baptist University", "ror": "https://ror.org/02dgbcj80", "country_code": "US", "type": "education", "lineage": ["https://openalex.org/I80804394"]}], "countries": ["HK", "US"], "is_corresponding": false, "raw_author_name": "Zhu, Jianing", "raw_affiliation_strings": ["TMLR Group , Department of Computer Science , Hong Kong Baptist University"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "TMLR Group , Department of Computer Science , Hong Kong Baptist University", "institution_ids": ["https://openalex.org/I141568987", "https://openalex.org/I80804394"]}]}, {"author_position": "middle", "author": {"id": null, "display_name": "Ge, Xinmu", "orcid": null, "observed_orcids": []}, "institutions": [{"id": "https://openalex.org/I183067930", "display_name": "Shanghai Jiao Tong University", "ror": "https://ror.org/0220qvk04", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I183067930"]}, {"id": "https://openalex.org/I4405279436", "display_name": "Shanghai Innovation Institute", "ror": "https://ror.org/01fatbw66", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I4405279436"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Ge, Xinmu", "raw_affiliation_strings": ["CMIC , Shanghai Jiao Tong University", "Shanghai Innovation Institute"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "CMIC , Shanghai Jiao Tong University", "institution_ids": ["https://openalex.org/I183067930"]}, {"raw_affiliation_string": "Shanghai Innovation Institute", "institution_ids": ["https://openalex.org/I4405279436"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5102705176", "display_name": "Zihua Zhao", "orcid": "https://orcid.org/0000-0001-8611-8656", "observed_orcids": ["https://orcid.org/0000-0001-8611-8656", "https://orcid.org/0009-0003-2915-6893"]}, "institutions": [{"id": "https://openalex.org/I183067930", "display_name": "Shanghai Jiao Tong University", "ror": "https://ror.org/0220qvk04", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I183067930"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Zhao, Zihua", "raw_affiliation_strings": ["CMIC , Shanghai Jiao Tong University"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "CMIC , Shanghai Jiao Tong University", "institution_ids": ["https://openalex.org/I183067930"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5101267097", "display_name": "Zhanke Zhou", "orcid": "https://orcid.org/0009-0008-1663-9358", "observed_orcids": ["https://orcid.org/0009-0008-1663-9358"]}, "institutions": [{"id": "https://openalex.org/I141568987", "display_name": "Hong Kong Baptist University", "ror": "https://ror.org/0145fw131", "country_code": "HK", "type": "education", "lineage": ["https://openalex.org/I141568987"]}, {"id": "https://openalex.org/I80804394", "display_name": "Southwest Baptist University", "ror": "https://ror.org/02dgbcj80", "country_code": "US", "type": "education", "lineage": ["https://openalex.org/I80804394"]}], "countries": ["HK", "US"], "is_corresponding": false, "raw_author_name": "Zhou, Zhanke", "raw_affiliation_strings": ["TMLR Group , Department of Computer Science , Hong Kong Baptist University"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "TMLR Group , Department of Computer Science , Hong Kong Baptist University", "institution_ids": ["https://openalex.org/I141568987", "https://openalex.org/I80804394"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5108047728", "display_name": "Xuan Li", "orcid": "https://orcid.org/0000-0002-9172-4105", "observed_orcids": ["https://orcid.org/0000-0002-9172-4105"]}, "institutions": [{"id": "https://openalex.org/I141568987", "display_name": "Hong Kong Baptist University", "ror": "https://ror.org/0145fw131", "country_code": "HK", "type": "education", "lineage": ["https://openalex.org/I141568987"]}, {"id": "https://openalex.org/I80804394", "display_name": "Southwest Baptist University", "ror": "https://ror.org/02dgbcj80", "country_code": "US", "type": "education", "lineage": ["https://openalex.org/I80804394"]}], "countries": ["HK", "US"], "is_corresponding": false, "raw_author_name": "Li, Xuan", "raw_affiliation_strings": ["TMLR Group , Department of Computer Science , Hong Kong Baptist University"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "TMLR Group , Department of Computer Science , Hong Kong Baptist University", "institution_ids": ["https://openalex.org/I141568987", "https://openalex.org/I80804394"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5003764862", "display_name": "Feng Xiao", "orcid": "https://orcid.org/0000-0002-2895-1727", "observed_orcids": ["https://orcid.org/0000-0002-2895-1727"]}, "institutions": [{"id": "https://openalex.org/I141568987", "display_name": "Hong Kong Baptist University", "ror": "https://ror.org/0145fw131", "country_code": "HK", "type": "education", "lineage": ["https://openalex.org/I141568987"]}, {"id": "https://openalex.org/I80804394", "display_name": "Southwest Baptist University", "ror": "https://ror.org/02dgbcj80", "country_code": "US", "type": "education", "lineage": ["https://openalex.org/I80804394"]}], "countries": ["HK", "US"], "is_corresponding": false, "raw_author_name": "Feng, Xiao", "raw_affiliation_strings": ["TMLR Group , Department of Computer Science , Hong Kong Baptist University"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "TMLR Group , Department of Computer Science , Hong Kong Baptist University", "institution_ids": ["https://openalex.org/I141568987", "https://openalex.org/I80804394"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5102922412", "display_name": "Jiangchao Yao", "orcid": "https://orcid.org/0000-0001-6115-5194", "observed_orcids": ["https://orcid.org/0000-0001-6115-5194"]}, "institutions": [{"id": "https://openalex.org/I183067930", "display_name": "Shanghai Jiao Tong University", "ror": "https://ror.org/0220qvk04", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I183067930"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Yao, Jiangchao", "raw_affiliation_strings": ["CMIC , Shanghai Jiao Tong University"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "CMIC , Shanghai Jiao Tong University", "institution_ids": ["https://openalex.org/I183067930"]}]}, {"author_position": "last", "author": {"id": "https://openalex.org/A5101418359", "display_name": "Bo Han", "orcid": "https://orcid.org/0000-0002-8633-7661", "observed_orcids": ["https://orcid.org/0000-0002-8633-7661"]}, "institutions": [{"id": "https://openalex.org/I141568987", "display_name": "Hong Kong Baptist University", "ror": "https://ror.org/0145fw131", "country_code": "HK", "type": "education", "lineage": ["https://openalex.org/I141568987"]}, {"id": "https://openalex.org/I80804394", "display_name": "Southwest Baptist University", "ror": "https://ror.org/02dgbcj80", "country_code": "US", "type": "education", "lineage": ["https://openalex.org/I80804394"]}], "countries": ["HK", "US"], "is_corresponding": false, "raw_author_name": "Han, Bo", "raw_affiliation_strings": ["TMLR Group , Department of Computer Science , Hong Kong Baptist University"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "TMLR Group , Department of Computer Science , Hong Kong Baptist University", "institution_ids": ["https://openalex.org/I141568987", "https://openalex.org/I80804394"]}]}], "abstract_inverted_index": {"While": [0], "reinforcement": [1], "learning": [2], "with": [3, 134, 206, 221], "verifiable": [4], "rewards": [5], "(RLVR)": [6], "is": [7, 106, 124, 229], "effective": [8], "to": [9, 25, 41, 137, 148, 166], "improve": [10], "the": [11, 26, 43, 52, 58, 64, 68, 73, 150, 169], "reasoning": [12, 44, 157, 192], "ability": [13], "of": [14, 46, 75, 146, 152, 217], "large": [15], "language": [16], "models": [17], "(LLMs),": [18], "its": [19], "reliance": [20], "on": [21, 155, 187, 189, 197, 219], "human-annotated": [22], "labels": [23, 136], "leads": [24], "scaling": [27], "up": [28], "dilemma,": [29], "especially": [30, 194], "for": [31], "complex": [32], "tasks.": [33], "Recent": [34], "self-rewarding": [35, 182], "methods": [36], "investigate": [37], "a": [38, 81, 107, 125, 130, 215], "label-free": [39], "alternative": [40], "unlock": [42], "capabilities": [45], "LLMs,": [47], "yet": [48], "they": [49], "frequently": [50], "encounter": [51], "non-negligible": [53], "training": [54, 88, 153, 175], "collapse": [55, 154], "issue,": [56], "as": [57, 214], "single-view": [59], "supervision": [60, 93], "signal": [61], "easily": [62], "forms": [63], "self-consistent": [65], "illusion,": [66], "yielding": [67], "reward": [69, 112], "hacking.": [70], "Inspired": [71], "by": [72, 90, 184, 195], "success": [74], "self-supervised": [76, 83], "learning,": [77], "we": [78, 98], "propose": [79], "\\textit{Co-rewarding},": [80], "novel": [82], "RL": [84], "framework": [85], "that": [86, 110, 128], "improves": [87], "stability": [89], "seeking": [91], "complementary": [92], "from": [94, 114], "another": [95], "views.": [96], "Specifically,": [97], "instantiate": [99], "Co-rewarding": [100, 172, 200], "in": [101, 210], "two": [102], "ways:": [103], "(1)": [104], "\\textit{Co-rewarding-I}": [105], "data-side": [108], "instantiation": [109, 127], "derives": [111], "signals": [113], "contrastive": [115], "agreement": [116], "across": [117, 176], "semantically": [118], "analogous": [119], "questions;": [120], "and": [121, 179], "(2)": [122], "\\textit{Co-rewarding-II}": [123], "model-side": [126], "maintains": [129], "slowly-updated": [131], "reference": [132], "teacher": [133], "pseudo": [135], "realize": [138], "self-distillation.": [139], "Intuitively,": [140], "such": [141, 213], "instantiations": [142], "introduce": [143], "different": [144], "levels": [145], "discrepancy": [147], "increase": [149], "difficulty": [151], "trivial": [156], "solutions.": [158], "We": [159], "also": [160], "explore": [161], "their": [162], "orthogonally": [163], "combined": [164], "version": [165], "further": [167], "boost": [168], "performance.": [170], "Empirically,": [171], "exhibits": [173], "stable": [174], "various": [177], "setups,": [178], "outperforms": [180], "other": [181], "baselines": [183], "$+3.31\\%$": [185], "improvements": [186], "average": [188], "multiple": [190], "mathematical": [191], "benchmarks,": [193], "$+7.49\\%$": [196], "Llama-3.2-3B-Instruct.": [198], "Notably,": [199], "reaches": [201], "or": [202], "even": [203], "surpasses": [204], "RLVR": [205], "ground-truth": [207], "(GT)": [208], "label": [209], "several": [211], "cases,": [212], "Pass@$1$": [216], "$94.01\\%$": [218], "GSM8K": [220], "Qwen3-8B-Base": [222], "remarkably": [223], "higher": [224], "than": [225], "GT.": [226], "Our": [227], "code": [228], "released": [230], "at": [231], "https://github.com/tmlr-group/Co-rewarding.": [232]}, "primary_topic": {"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.5985999703407288, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.5985999703407288, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10181", "display_name": "Natural Language Processing Techniques", "score": 0.08420000225305557, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T11307", "display_name": "Domain Adaptation and Few-Shot Learning", "score": 0.05700000002980232, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "pmh:oai:arXiv.org:2508.00410", "is_oa": true, "landing_page_url": "http://arxiv.org/abs/2508.00410", "pdf_url": "https://arxiv.org/pdf/2508.00410", "source": {"id": "https://openalex.org/S4306400194", "display_name": "arXiv (Cornell University)", "issn_l": "2331-8422", "issn": ["2331-8422"], "is_oa": true, "is_in_doaj": false, "is_core": false, "listed_in": [], "host_organization": "https://openalex.org/I205783295", "host_organization_name": "Cornell University", "host_organization_lineage": ["https://openalex.org/I205783295"], "host_organization_lineage_names": [], "type": "repository"}, "license": "cc-by", "license_id": "https://openalex.org/licenses/cc-by", "version": "submittedVersion", "is_accepted": false, "is_published": false, "raw_source_name": null, "raw_type": "text"}, "cited_by_count": 0, "is_retracted": false}, {"id": "https://openalex.org/W4412661742", "doi": "https://doi.org/10.1016/j.neucom.2025.131116", "title": "AcuGPT-Agent: An LLM-powered intelligent system for acupuncture-based infertility treatment", "publication_year": 2025, "publication_date": "2025-07-25", "type": "article", "language": "en", "authorships": [{"author_position": "first", "author": {"id": "https://openalex.org/A5102616782", "display_name": "Jing Wen", "orcid": "https://orcid.org/0009-0002-3376-5678", "observed_orcids": ["https://orcid.org/0009-0002-3376-5678"]}, "institutions": [{"id": "https://openalex.org/I131434179", "display_name": "Nanjing University of Chinese Medicine", "ror": "https://ror.org/04523zj19", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I131434179"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Jing Wen", "raw_affiliation_strings": ["College of Acupuncture Moxibustion and Tuina, Nanjing University of Chinese Medicine, Nanjing, 210023, PR China"], "raw_orcid": "https://orcid.org/0009-0002-3376-5678", "affiliations": [{"raw_affiliation_string": "College of Acupuncture Moxibustion and Tuina, Nanjing University of Chinese Medicine, Nanjing, 210023, PR China", "institution_ids": ["https://openalex.org/I131434179"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5017211268", "display_name": "Diandong Liu", "orcid": "https://orcid.org/0009-0003-7094-2511", "observed_orcids": ["https://orcid.org/0009-0003-7094-2511"]}, "institutions": [{"id": "https://openalex.org/I51622183", "display_name": "Shaanxi University of Science and Technology", "ror": "https://ror.org/034t3zs45", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I51622183"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Diandong Liu", "raw_affiliation_strings": ["Shaanxi Joint Laboratory of Artificial Intelligence, Shaanxi University of Science and Technology, Xi\u2019an, 710021, PR China"], "raw_orcid": "https://orcid.org/0009-0003-7094-2511", "affiliations": [{"raw_affiliation_string": "Shaanxi Joint Laboratory of Artificial Intelligence, Shaanxi University of Science and Technology, Xi\u2019an, 710021, PR China", "institution_ids": ["https://openalex.org/I51622183"]}]}, {"author_position": "middle", "author": {"id": null, "display_name": "Yuxin Xie", "orcid": "https://orcid.org/0009-0003-5709-8360", "observed_orcids": ["https://orcid.org/0009-0003-5709-8360"]}, "institutions": [{"id": "https://openalex.org/I4210111013", "display_name": "Nanjing Drum Tower Hospital", "ror": "https://ror.org/026axqv54", "country_code": "CN", "type": "healthcare", "lineage": ["https://openalex.org/I4210111013"]}, {"id": "https://openalex.org/I881766915", "display_name": "Nanjing University", "ror": "https://ror.org/01rxvg760", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I881766915"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Yuxin Xie", "raw_affiliation_strings": ["Department of Neurosurgery, Nanjing Drum Tower Hospital, Medical School of Nanjing University, Nanjing, 211166, PR China"], "raw_orcid": "https://orcid.org/0009-0003-5709-8360", "affiliations": [{"raw_affiliation_string": "Department of Neurosurgery, Nanjing Drum Tower Hospital, Medical School of Nanjing University, Nanjing, 211166, PR China", "institution_ids": ["https://openalex.org/I4210111013", "https://openalex.org/I881766915"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5119075863", "display_name": "Yi Ren", "orcid": "https://orcid.org/0009-0008-8453-9447", "observed_orcids": ["https://orcid.org/0009-0008-8453-9447"]}, "institutions": [{"id": "https://openalex.org/I881766915", "display_name": "Nanjing University", "ror": "https://ror.org/01rxvg760", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I881766915"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Yi Ren", "raw_affiliation_strings": ["School of Software Engineering, Nanjing University, Nanjing, 210093, PR China"], "raw_orcid": "https://orcid.org/0009-0008-8453-9447", "affiliations": [{"raw_affiliation_string": "School of Software Engineering, Nanjing University, Nanjing, 210093, PR China", "institution_ids": ["https://openalex.org/I881766915"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5069580705", "display_name": "Jiacun Wang", "orcid": "https://orcid.org/0000-0003-4176-3947", "observed_orcids": ["https://orcid.org/0000-0003-4176-3947"]}, "institutions": [{"id": "https://openalex.org/I141450709", "display_name": "Monmouth University", "ror": "https://ror.org/01d6qxv05", "country_code": "US", "type": "education", "lineage": ["https://openalex.org/I141450709"]}], "countries": ["US"], "is_corresponding": true, "raw_author_name": "Jiacun Wang", "raw_affiliation_strings": ["Department of Computer Science and Software Engineering, Monmouth University, W.Long Branch, NJ 07764, USA"], "raw_orcid": "https://orcid.org/0000-0003-4176-3947", "affiliations": [{"raw_affiliation_string": "Department of Computer Science and Software Engineering, Monmouth University, W.Long Branch, NJ 07764, USA", "institution_ids": ["https://openalex.org/I141450709"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5101717171", "display_name": "Youbing Xia", "orcid": "https://orcid.org/0000-0002-0408-2045", "observed_orcids": ["https://orcid.org/0000-0002-0408-2045", "https://orcid.org/0009-0008-4476-4738"]}, "institutions": [{"id": "https://openalex.org/I131434179", "display_name": "Nanjing University of Chinese Medicine", "ror": "https://ror.org/04523zj19", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I131434179"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Youbing Xia", "raw_affiliation_strings": ["College of Acupuncture Moxibustion and Tuina, Nanjing University of Chinese Medicine, Nanjing, 210023, PR China"], "raw_orcid": "https://orcid.org/0000-0002-0408-2045", "affiliations": [{"raw_affiliation_string": "College of Acupuncture Moxibustion and Tuina, Nanjing University of Chinese Medicine, Nanjing, 210023, PR China", "institution_ids": ["https://openalex.org/I131434179"]}]}, {"author_position": "last", "author": {"id": "https://openalex.org/A5023758419", "display_name": "Peng Zhu", "orcid": "https://orcid.org/0000-0003-3687-9187", "observed_orcids": ["https://orcid.org/0000-0003-3687-9187"]}, "institutions": [{"id": "https://openalex.org/I36399199", "display_name": "Nanjing University of Science and Technology", "ror": "https://ror.org/00xp9wg62", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I36399199"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Peng Zhu", "raw_affiliation_strings": ["School of Economics and Management, Nanjing University of Science and Technology, Nanjing, 210094, PR China"], "raw_orcid": "https://orcid.org/0000-0003-3687-9187", "affiliations": [{"raw_affiliation_string": "School of Economics and Management, Nanjing University of Science and Technology, Nanjing, 210094, PR China", "institution_ids": ["https://openalex.org/I36399199"]}]}], "abstract_inverted_index": null, "primary_topic": {"id": "https://openalex.org/T11636", "display_name": "Artificial Intelligence in Healthcare and Education", "score": 0.7621999979019165, "subfield": {"id": "https://openalex.org/subfields/2718", "display_name": "Health Informatics"}, "field": {"id": "https://openalex.org/fields/27", "display_name": "Medicine"}, "domain": {"id": "https://openalex.org/domains/4", "display_name": "Health Sciences"}}, "topics": [{"id": "https://openalex.org/T11636", "display_name": "Artificial Intelligence in Healthcare and Education", "score": 0.7621999979019165, "subfield": {"id": "https://openalex.org/subfields/2718", "display_name": "Health Informatics"}, "field": {"id": "https://openalex.org/fields/27", "display_name": "Medicine"}, "domain": {"id": "https://openalex.org/domains/4", "display_name": "Health Sciences"}}, {"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.11999999731779099, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10286", "display_name": "Information Retrieval and Search Behavior", "score": 0.01269999984651804, "subfield": {"id": "https://openalex.org/subfields/1710", "display_name": "Information Systems"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "doi:10.1016/j.neucom.2025.131116", "is_oa": false, "landing_page_url": "https://doi.org/10.1016/j.neucom.2025.131116", "pdf_url": null, "source": {"id": "https://openalex.org/S45693802", "display_name": "Neurocomputing", "issn_l": "0925-2312", "issn": ["0925-2312", "1872-8286"], "is_oa": false, "is_in_doaj": false, "is_core": true, "listed_in": ["cwts-core", "doyens", "jufo-2", "ki-jl-2", "norway-2"], "host_organization": "https://openalex.org/P4310320990", "host_organization_name": "Elsevier BV", "host_organization_lineage": ["https://openalex.org/P4310320990"], "host_organization_lineage_names": ["Elsevier BV"], "type": "journal"}, "license": null, "license_id": null, "version": "publishedVersion", "is_accepted": true, "is_published": true, "raw_source_name": "Neurocomputing", "raw_type": "journal-article"}, "cited_by_count": 9, "is_retracted": false}, {"id": "https://openalex.org/W4414244021", "doi": "https://doi.org/10.3390/s25185758", "title": "KG-SR-LLM: Knowledge-Guided Semantic Representation and Large Language Model Framework for Cross-Domain Bearing Fault Diagnosis", "publication_year": 2025, "publication_date": "2025-09-16", "type": "article", "language": "en", "authorships": [{"author_position": "first", "author": {"id": "https://openalex.org/A5064653166", "display_name": "Chengyong Xiao", "orcid": null, "observed_orcids": []}, "institutions": [{"id": "https://openalex.org/I92403157", "display_name": "University of Science and Technology Beijing", "ror": "https://ror.org/02egmk993", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I92403157"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Chengyong Xiao", "raw_affiliation_strings": ["School of Automation & Electrical Engineering, University of Science and Technology Beijing, Beijing 100083, China"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "School of Automation & Electrical Engineering, University of Science and Technology Beijing, Beijing 100083, China", "institution_ids": ["https://openalex.org/I92403157"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5100390771", "display_name": "Xiaowei Liu", "orcid": "https://orcid.org/0000-0003-2074-3760", "observed_orcids": ["https://orcid.org/0000-0003-2074-3760"]}, "institutions": [{"id": "https://openalex.org/I92403157", "display_name": "University of Science and Technology Beijing", "ror": "https://ror.org/02egmk993", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I92403157"]}], "countries": ["CN"], "is_corresponding": true, "raw_author_name": "Xiaowei Liu", "raw_affiliation_strings": ["School of Computer & Communication Engineering, University of Science and Technology Beijing, Beijing 100083, China"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "School of Computer & Communication Engineering, University of Science and Technology Beijing, Beijing 100083, China", "institution_ids": ["https://openalex.org/I92403157"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5086313562", "display_name": "Aziguli Wulamu", "orcid": "https://orcid.org/0000-0001-7228-7838", "observed_orcids": ["https://orcid.org/0000-0001-7228-7838", "https://orcid.org/0000-0002-3320-8110"]}, "institutions": [{"id": "https://openalex.org/I92403157", "display_name": "University of Science and Technology Beijing", "ror": "https://ror.org/02egmk993", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I92403157"]}], "countries": ["CN"], "is_corresponding": true, "raw_author_name": "Aziguli Wulamu", "raw_affiliation_strings": ["School of Computer & Communication Engineering, University of Science and Technology Beijing, Beijing 100083, China"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "School of Computer & Communication Engineering, University of Science and Technology Beijing, Beijing 100083, China", "institution_ids": ["https://openalex.org/I92403157"]}]}, {"author_position": "last", "author": {"id": "https://openalex.org/A5024869142", "display_name": "Dezheng Zhang", "orcid": "https://orcid.org/0000-0002-3456-5259", "observed_orcids": ["https://orcid.org/0000-0002-3456-5259"]}, "institutions": [{"id": "https://openalex.org/I92403157", "display_name": "University of Science and Technology Beijing", "ror": "https://ror.org/02egmk993", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I92403157"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Dezheng Zhang", "raw_affiliation_strings": ["School of Computer & Communication Engineering, University of Science and Technology Beijing, Beijing 100083, China"], "raw_orcid": "https://orcid.org/0000-0002-3456-5259", "affiliations": [{"raw_affiliation_string": "School of Computer & Communication Engineering, University of Science and Technology Beijing, Beijing 100083, China", "institution_ids": ["https://openalex.org/I92403157"]}]}], "abstract_inverted_index": {"Bearing": [0], "fault": [1, 139, 154, 239], "diagnosis": [2, 155, 240], "is": [3, 70, 126, 166, 198, 215], "crucial": [4], "for": [5, 217, 237], "stable": [6], "operation": [7], "and": [8, 23, 63, 90, 102, 112, 161, 177, 192, 221, 232], "safe": [9], "manufacturing": [10], "as": [11], "industry": [12], "intelligence": [13], "becomes": [14], "increasingly": [15], "advanced.": [16], "However,": [17], "under": [18], "complicated": [19], "non-linear": [20], "vibration": [21, 75], "modes": [22], "multiple": [24, 57], "operating": [25, 113], "conditions,": [26], "most": [27], "of": [28, 37, 195, 212, 235, 241], "the": [29, 96, 131, 148, 229], "current": [30], "diagnostic": [31, 49, 210], "methods": [32], "are": [33, 180], "limited": [34], "in": [35, 87], "terms": [36], "cross-domain": [38], "generalization.": [39], "To": [40], "address": [41], "these": [42, 226], "issues,": [43], "this": [44], "study": [45], "develops": [46], "a": [47, 66, 116, 152, 193], "generalized": [48, 153], "framework": [50], "leveraging": [51], "Large": [52, 162], "Language": [53, 163], "Models": [54], "(LLMs),": [55], "integrating": [56], "enhancements": [58], "to": [59, 72, 106, 133, 137, 183, 200], "improve": [60], "both": [61, 88], "accuracy": [62, 211], "adaptability.": [64], "Initially,": [65], "structured": [67], "representation": [68], "approach": [69], "designed": [71], "transform": [73], "raw": [74], "time": [76, 89], "series": [77], "into": [78, 147], "interpretable": [79], "text": [80], "sequences": [81], "by": [82, 141, 205], "extracting": [83], "physically": [84], "meaningful": [85], "features": [86], "frequency": [91], "domains.": [92], "This": [93, 128], "transformation": [94], "bridges": [95], "gap": [97], "between": [98], "sequential": [99], "sensor": [100], "data": [101], "semantic": [103], "understanding.": [104], "Furthermore,": [105], "explicitly": [107], "incorporate": [108], "bearings'": [109], "structural": [110], "parameters": [111], "condition": [114], "information,": [115], "knowledge-guided": [117], "prompt": [118], "tuning": [119], "strategy": [120], "based": [121], "on": [122, 189], "Low-Rank": [123], "Adaptation": [124], "(LoRA-Prompt)": [125], "introduced.": [127], "mechanism": [129], "enables": [130], "model": [132], "adapt": [134], "more": [135], "effectively": [136], "varying": [138], "scenarios": [140], "embedding": [142], "expert": [143], "prior": [144], "knowledge": [145], "directly": [146], "learning": [149, 203], "process.": [150], "Finally,": [151], "method": [156], "named": [157], "Knowledge-Guided": [158], "Semantic": [159], "Representation": [160], "Model": [164], "(KG-SR-LLM)": [165], "established.": [167], "Large-scale": [168], "experiments": [169], "using": [170], "11": [171], "public": [172], "datasets": [173], "from": [174], "industrial,": [175], "aerospace,": [176], "energy": [178], "fields": [179], "carried": [181], "out": [182], "extensively": [184], "evaluate": [185], "its": [186], "performance.": [187], "Based": [188], "experiment": [190], "analysis": [191], "comparison": [194], "results,": [196], "KG-SR-LLM": [197, 214, 236], "superior": [199], "classical": [201], "deep": [202], "models": [204], "9.22%,": [206], "reaching": [207], "an": [208], "average": [209], "98.36%.": [213], "effective": [216], "handling": [218], "few-shot": [219], "transfer": [220], "cross-condition": [222], "adaptation": [223], "tasks.": [224], "All": [225], "results": [227], "illustrate": [228], "theoretical": [230], "significance": [231], "application": [233], "benefit": [234], "intelligent": [238], "bearings.": [242]}, "primary_topic": {"id": "https://openalex.org/T10220", "display_name": "Machine Fault Diagnosis Techniques", "score": 1.0, "subfield": {"id": "https://openalex.org/subfields/2207", "display_name": "Control and Systems Engineering"}, "field": {"id": "https://openalex.org/fields/22", "display_name": "Engineering"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T10220", "display_name": "Machine Fault Diagnosis Techniques", "score": 1.0, "subfield": {"id": "https://openalex.org/subfields/2207", "display_name": "Control and Systems Engineering"}, "field": {"id": "https://openalex.org/fields/22", "display_name": "Engineering"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10876", "display_name": "Fault Detection and Control Systems", "score": 0.0, "subfield": {"id": "https://openalex.org/subfields/2207", "display_name": "Control and Systems Engineering"}, "field": {"id": "https://openalex.org/fields/22", "display_name": "Engineering"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T11062", "display_name": "Gear and Bearing Dynamics Analysis", "score": 0.0, "subfield": {"id": "https://openalex.org/subfields/2210", "display_name": "Mechanical Engineering"}, "field": {"id": "https://openalex.org/fields/22", "display_name": "Engineering"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "doi:10.3390/s25185758", "is_oa": true, "landing_page_url": "https://doi.org/10.3390/s25185758", "pdf_url": "https://www.mdpi.com/1424-8220/25/18/5758/pdf?version=1758017114", "source": {"id": "https://openalex.org/S101949793", "display_name": "Sensors", "issn_l": "1424-8220", "issn": ["1424-8220"], "is_oa": true, "is_in_doaj": true, "is_core": true, "listed_in": ["cwts-core", "doaj", "jufo-1", "medline", "norway-1"], "host_organization": "https://openalex.org/P4310310987", "host_organization_name": "Multidisciplinary Digital Publishing Institute", "host_organization_lineage": ["https://openalex.org/P4310310987"], "host_organization_lineage_names": ["Multidisciplinary Digital Publishing Institute"], "type": "journal"}, "license": "cc-by", "license_id": "https://openalex.org/licenses/cc-by", "version": "publishedVersion", "is_accepted": true, "is_published": true, "raw_source_name": "Sensors", "raw_type": "journal-article"}, "cited_by_count": 13, "is_retracted": false}, {"id": "https://openalex.org/W4414836283", "doi": "https://doi.org/10.48550/arxiv.2502.15920", "title": "Self-Taught Agentic Long Context Understanding", "publication_year": 2025, "publication_date": "2025-02-21", "type": "preprint", "language": "en", "authorships": [{"author_position": "first", "author": {"id": "https://openalex.org/A5052671723", "display_name": "Yufan Zhuang", "orcid": "https://orcid.org/0000-0003-4063-6460", "observed_orcids": ["https://orcid.org/0000-0003-4063-6460", "https://orcid.org/0009-0001-7506-7475"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Zhuang, Yufan", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5139762370", "display_name": "Xiaodong Yu", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Yu, Xiaodong", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5115829690", "display_name": "Jialian Wu", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Wu, Jialian", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5101181950", "display_name": "Ximeng Sun", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Sun, Ximeng", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5107953191", "display_name": "Ze Wang", "orcid": "https://orcid.org/0009-0000-4619-7943", "observed_orcids": ["https://orcid.org/0009-0000-4619-7943"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Wang, Ze", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5100325209", "display_name": "Jiang Liu", "orcid": "https://orcid.org/0000-0001-6281-6505", "observed_orcids": ["https://orcid.org/0000-0001-6281-6505"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Liu, Jiang", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5011372688", "display_name": "Yu-Sheng Su", "orcid": "https://orcid.org/0000-0002-1531-3363", "observed_orcids": ["https://orcid.org/0000-0002-1531-3363"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Su, Yusheng", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5039500313", "display_name": "Jingbo Shang", "orcid": "https://orcid.org/0000-0002-7249-4404", "observed_orcids": ["https://orcid.org/0000-0002-7249-4404"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Shang, Jingbo", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5101728120", "display_name": "Zicheng Liu", "orcid": "https://orcid.org/0000-0003-1106-2963", "observed_orcids": ["https://orcid.org/0000-0003-1106-2963"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Liu, Zicheng", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "last", "author": {"id": "https://openalex.org/A5027115167", "display_name": "Emad Barsoum", "orcid": "https://orcid.org/0000-0002-4097-8690", "observed_orcids": ["https://orcid.org/0000-0002-4097-8690"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Barsoum, Emad", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}], "abstract_inverted_index": {"Answering": [0], "complex,": [1], "long-context": [2, 175, 187], "questions": [3, 66], "remains": [4], "a": [5, 28, 75, 82, 93, 101, 167], "major": [6], "challenge": [7], "for": [8, 122], "large": [9], "language": [10], "models": [11, 59, 155], "(LLMs)": [12], "as": [13, 74, 197], "it": [14], "requires": [15], "effective": [16, 140], "question": [17], "clarifications": [18, 158], "and": [19, 67, 100, 130, 143, 159, 164, 185], "context": [20, 162, 198], "retrieval.": [21], "We": [22], "propose": [23], "Agentic": [24], "Long-Context": [25], "Understanding": [26], "(AgenticLU),": [27], "framework": [29], "designed": [30], "to": [31, 98, 115, 138, 148, 156], "enhance": [32, 149], "an": [33, 47], "LLM's": [34], "understanding": [35, 62], "of": [36, 53, 96, 104, 111], "such": [37], "queries": [38], "by": [39, 126], "integrating": [40], "targeted": [41], "self-clarification": [42], "with": [43, 92], "contextual": [44, 69], "grounding": [45], "within": [46], "agentic": [48], "workflow.": [49], "At": [50], "the": [51, 108, 119, 127], "core": [52], "AgenticLU": [54, 154, 179], "is": [55], "Chain-of-Clarifications": [56], "(CoC),": [57], "where": [58, 78], "refine": [60], "their": [61], "through": [63], "self-generated": [64], "clarification": [65], "corresponding": [68], "groundings.": [70], "By": [71], "scaling": [72], "inference": [73, 169], "tree": [76], "search": [77, 94, 113], "each": [79, 123], "node": [80], "represents": [81], "CoC": [83, 128], "step,": [84], "we": [85, 117], "achieve": [86], "97.8%": [87], "answer": [88], "recall": [89], "on": [90], "NarrativeQA": [91], "depth": [95], "up": [97], "three": [99], "branching": [102], "factor": [103], "eight.": [105], "To": [106], "amortize": [107], "high": [109], "cost": [110], "this": [112], "process": [114], "training,": [116], "leverage": [118], "preference": [120, 146], "pairs": [121], "step": [124], "obtained": [125], "workflow": [129], "perform": [131], "two-stage": [132], "model": [133], "finetuning:": [134], "(1)": [135], "supervised": [136], "finetuning": [137], "learn": [139], "decomposition": [141], "strategies,": [142], "(2)": [144], "direct": [145], "optimization": [147], "reasoning": [150, 192], "quality.": [151], "This": [152], "enables": [153], "generate": [157], "retrieve": [160], "relevant": [161], "effectively": [163], "efficiently": [165], "in": [166], "single": [168], "pass.": [170], "Extensive": [171], "experiments": [172], "across": [173], "seven": [174], "tasks": [176], "demonstrate": [177], "that": [178], "significantly": [180], "outperforms": [181], "state-of-the-art": [182], "prompting": [183], "methods": [184], "specialized": [186], "LLMs,": [188], "achieving": [189], "robust": [190], "multi-hop": [191], "while": [193], "sustaining": [194], "consistent": [195], "performance": [196], "length": [199], "grows.": [200]}, "primary_topic": {"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.9404000043869019, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.9404000043869019, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T11714", "display_name": "Multimodal Machine Learning Applications", "score": 0.025599999353289604, "subfield": {"id": "https://openalex.org/subfields/1707", "display_name": "Computer Vision and Pattern Recognition"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10181", "display_name": "Natural Language Processing Techniques", "score": 0.01600000075995922, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "pmh:oai:arXiv.org:2502.15920", "is_oa": true, "landing_page_url": "http://arxiv.org/abs/2502.15920", "pdf_url": "https://arxiv.org/pdf/2502.15920", "source": {"id": "https://openalex.org/S4306400194", "display_name": "arXiv (Cornell University)", "issn_l": "2331-8422", "issn": ["2331-8422"], "is_oa": true, "is_in_doaj": false, "is_core": false, "listed_in": [], "host_organization": "https://openalex.org/I205783295", "host_organization_name": "Cornell University", "host_organization_lineage": ["https://openalex.org/I205783295"], "host_organization_lineage_names": [], "type": "repository"}, "license": null, "license_id": null, "version": "submittedVersion", "is_accepted": false, "is_published": false, "raw_source_name": null, "raw_type": "text"}, "cited_by_count": 0, "is_retracted": false}, {"id": "https://openalex.org/W7117130051", "doi": "https://doi.org/10.3390/ijgi15010008", "title": "Domain-Adapted MLLMs for Interpretable Road Traffic Accident Analysis Using Remote Sensing Imagery", "publication_year": 2025, "publication_date": "2025-12-21", "type": "article", "language": "en", "authorships": [{"author_position": "first", "author": {"id": "https://openalex.org/A5113347971", "display_name": "Bing He", "orcid": "https://orcid.org/0009-0002-2510-7069", "observed_orcids": ["https://orcid.org/0009-0002-2510-7069"]}, "institutions": [{"id": "https://openalex.org/I24201400", "display_name": "Chengdu University of Information Technology", "ror": "https://ror.org/01yxwrh59", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I24201400"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Bing He", "raw_affiliation_strings": ["School of Computer Science, Chengdu University of Information Technology, Chengdu 610225, China"], "raw_orcid": "https://orcid.org/0009-0002-2510-7069", "affiliations": [{"raw_affiliation_string": "School of Computer Science, Chengdu University of Information Technology, Chengdu 610225, China", "institution_ids": ["https://openalex.org/I24201400"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5121133911", "display_name": "Wei He", "orcid": null, "observed_orcids": []}, "institutions": [{"id": "https://openalex.org/I24201400", "display_name": "Chengdu University of Information Technology", "ror": "https://ror.org/01yxwrh59", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I24201400"]}], "countries": ["CN"], "is_corresponding": true, "raw_author_name": "Wei He", "raw_affiliation_strings": ["School of Computer Science, Chengdu University of Information Technology, Chengdu 610225, China"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "School of Computer Science, Chengdu University of Information Technology, Chengdu 610225, China", "institution_ids": ["https://openalex.org/I24201400"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5121217546", "display_name": "Qing Chang", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Qing Chang", "raw_affiliation_strings": ["Sain Associates, Inc., 5021 Technology Drive Northwest, Suite B2, Huntsville, AL 35805, USA"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Sain Associates, Inc., 5021 Technology Drive Northwest, Suite B2, Huntsville, AL 35805, USA", "institution_ids": []}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5121229142", "display_name": "Wen Luo", "orcid": "https://orcid.org/0009-0004-1409-4738", "observed_orcids": ["https://orcid.org/0009-0004-1409-4738"]}, "institutions": [{"id": "https://openalex.org/I24201400", "display_name": "Chengdu University of Information Technology", "ror": "https://ror.org/01yxwrh59", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I24201400"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Wen Luo", "raw_affiliation_strings": ["School of Computer Science, Chengdu University of Information Technology, Chengdu 610225, China"], "raw_orcid": "https://orcid.org/0009-0004-1409-4738", "affiliations": [{"raw_affiliation_string": "School of Computer Science, Chengdu University of Information Technology, Chengdu 610225, China", "institution_ids": ["https://openalex.org/I24201400"]}]}, {"author_position": "last", "author": {"id": "https://openalex.org/A5121189741", "display_name": "Lingli Xiao", "orcid": null, "observed_orcids": []}, "institutions": [{"id": "https://openalex.org/I24201400", "display_name": "Chengdu University of Information Technology", "ror": "https://ror.org/01yxwrh59", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I24201400"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Lingli Xiao", "raw_affiliation_strings": ["School of Computer Science, Chengdu University of Information Technology, Chengdu 610225, China"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "School of Computer Science, Chengdu University of Information Technology, Chengdu 610225, China", "institution_ids": ["https://openalex.org/I24201400"]}]}], "abstract_inverted_index": {"Traditional": [0], "road": [1, 38], "traffic": [2, 39, 192], "accident": [3, 31, 40, 58, 88, 106, 138], "analysis": [4, 41, 189], "has": [5], "long": [6], "relied": [7], "on": [8, 44], "structured": [9, 57, 172], "data,": [10], "making": [11], "it": [12], "difficult": [13], "to": [14, 26, 77], "integrate": [15], "high-dimensional": [16], "heterogeneous": [17], "information": [18], "such": [19], "as": [20], "remote": [21, 53, 102], "sensing": [22, 54, 103], "imagery": [23, 55], "and": [24, 90, 109, 140, 149, 161, 187], "leading": [25], "an": [27, 145, 150], "incomplete": [28], "understanding": [29], "of": [30, 137, 147, 152, 171], "scene": [32], "environments.": [33], "This": [34], "study": [35], "proposes": [36], "a": [37, 61, 121, 182], "framework": [42], "based": [43], "Multimodal": [45], "Large": [46], "Language": [47], "Models.": [48], "The": [49], "approach": [50], "integrates": [51], "high-resolution": [52], "with": [56], "data": [59], "through": [60], "three-stage": [62], "progressive": [63], "training": [64], "pipeline.": [65], "Specifically,": [66], "we": [67], "fine-tune": [68], "three": [69], "open-source": [70], "vision\u2013language": [71], "models": [72], "using": [73], "Low-Rank": [74], "Adaptation": [75], "(LoRA)": [76], "sequentially": [78], "optimize": [79], "the": [80, 117, 125, 133, 142, 155, 169], "model\u2019s": [81], "capabilities": [82], "in": [83, 124, 175], "visual": [84], "environmental": [85], "description,": [86], "multi-task": [87], "classification,": [89], "Chain-of-Thought": [91], "(CoT)": [92], "driven": [93], "causal": [94], "reasoning.": [95], "A": [96], "multimodal": [97, 176], "dataset": [98], "was": [99], "constructed": [100], "containing": [101], "image": [104, 129], "descriptions,": [105], "classification": [107, 135], "labels,": [108], "interpretable": [110, 186], "reasoning": [111, 157, 173], "chains.": [112], "Experimental": [113], "results": [114, 167], "show": [115], "that": [116], "fine-tuned": [118], "model": [119, 143], "achieved": [120, 144], "maximum": [122], "improvement": [123], "CIDEr": [126, 162], "score": [127], "for": [128, 178, 190], "description": [130], "tasks.": [131], "In": [132, 154], "joint": [134], "task": [136], "severity": [139], "duration,": [141], "accuracy": [146], "71.61%": [148], "F1-score": [151], "0.8473.": [153], "CoT": [156], "task,": [158], "both": [159], "METEOR": [160], "scores": [163], "improved": [164], "significantly.": [165], "These": [166], "validate": [168], "effectiveness": [170], "mechanisms": [174], "fusion": [177], "transportation": [179], "applications,": [180], "providing": [181], "feasible": [183], "path": [184], "toward": [185], "intelligent": [188], "real-world": [191], "management.": [193]}, "primary_topic": {"id": "https://openalex.org/T11714", "display_name": "Multimodal Machine Learning Applications", "score": 0.9086999893188477, "subfield": {"id": "https://openalex.org/subfields/1707", "display_name": "Computer Vision and Pattern Recognition"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T11714", "display_name": "Multimodal Machine Learning Applications", "score": 0.9086999893188477, "subfield": {"id": "https://openalex.org/subfields/1707", "display_name": "Computer Vision and Pattern Recognition"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10370", "display_name": "Traffic and Road Safety", "score": 0.07029999792575836, "subfield": {"id": "https://openalex.org/subfields/2213", "display_name": "Safety, Risk, Reliability and Quality"}, "field": {"id": "https://openalex.org/fields/22", "display_name": "Engineering"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10036", "display_name": "Advanced Neural Network Applications", "score": 0.006000000052154064, "subfield": {"id": "https://openalex.org/subfields/1707", "display_name": "Computer Vision and Pattern Recognition"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "doi:10.3390/ijgi15010008", "is_oa": true, "landing_page_url": "https://doi.org/10.3390/ijgi15010008", "pdf_url": null, "source": {"id": "https://openalex.org/S2764431341", "display_name": "ISPRS International Journal of Geo-Information", "issn_l": "2220-9964", "issn": ["2220-9964"], "is_oa": true, "is_in_doaj": true, "is_core": true, "listed_in": ["cwts-core", "doaj", "jufo-1", "ki-jl-1", "norway-1"], "host_organization": "https://openalex.org/P4310310987", "host_organization_name": "Multidisciplinary Digital Publishing Institute", "host_organization_lineage": ["https://openalex.org/P4310310987"], "host_organization_lineage_names": ["Multidisciplinary Digital Publishing Institute"], "type": "journal"}, "license": "cc-by", "license_id": "https://openalex.org/licenses/cc-by", "version": "publishedVersion", "is_accepted": true, "is_published": true, "raw_source_name": "ISPRS International Journal of Geo-Information", "raw_type": "journal-article"}, "cited_by_count": 2, "is_retracted": false}, {"id": "https://openalex.org/W4416609205", "doi": "https://doi.org/10.48550/arxiv.2505.23094", "title": "MAP: Revisiting Weight Decomposition for Low-Rank Adaptation", "publication_year": 2025, "publication_date": "2025-05-29", "type": "preprint", "language": "en", "authorships": [{"author_position": "first", "author": {"id": "https://openalex.org/A5100493819", "display_name": "Chongjie Si", "orcid": "https://orcid.org/0009-0001-3017-7574", "observed_orcids": ["https://orcid.org/0009-0001-3017-7574"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Si, Chongjie", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": null, "display_name": "Shi, Zhiyi", "orcid": null, "observed_orcids": []}, "institutions": [{"id": "https://openalex.org/I136199984", "display_name": "Harvard University", "ror": "https://ror.org/03vek6s52", "country_code": "US", "type": "education", "lineage": ["https://openalex.org/I136199984"]}], "countries": ["US"], "is_corresponding": false, "raw_author_name": "Shi, Zhiyi", "raw_affiliation_strings": ["Harvard University"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Harvard University", "institution_ids": ["https://openalex.org/I136199984"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5006309518", "display_name": "Yadao Wang", "orcid": null, "observed_orcids": []}, "institutions": [{"id": "https://openalex.org/I45928872", "display_name": "Alibaba Group (China)", "ror": "https://ror.org/00k642b80", "country_code": "CN", "type": "company", "lineage": ["https://openalex.org/I45928872"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Wang, Yadao", "raw_affiliation_strings": ["Alibaba Group ,"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Alibaba Group ,", "institution_ids": ["https://openalex.org/I45928872"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5019708391", "display_name": "Xiaokang Yang", "orcid": "https://orcid.org/0000-0003-4029-3322", "observed_orcids": ["https://orcid.org/0000-0003-4029-3322"]}, "institutions": [{"id": "https://openalex.org/I183067930", "display_name": "Shanghai Jiao Tong University", "ror": "https://ror.org/0220qvk04", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I183067930"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Yang, Xiaokang", "raw_affiliation_strings": ["Shanghai Jiao Tong University ,"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Shanghai Jiao Tong University ,", "institution_ids": ["https://openalex.org/I183067930"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5054558832", "display_name": "Susanto Rahardja", "orcid": "https://orcid.org/0000-0003-0831-6934", "observed_orcids": ["https://orcid.org/0000-0003-0831-6934"]}, "institutions": [{"id": "https://openalex.org/I168639165", "display_name": "Singapore Institute of Technology", "ror": "https://ror.org/01v2c2791", "country_code": "SG", "type": "education", "lineage": ["https://openalex.org/I168639165"]}], "countries": ["SG"], "is_corresponding": false, "raw_author_name": "Rahardja, Susanto", "raw_affiliation_strings": ["Singapore Institute of Technology"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Singapore Institute of Technology", "institution_ids": ["https://openalex.org/I168639165"]}]}, {"author_position": "last", "author": {"id": null, "display_name": "Shen, Wei", "orcid": null, "observed_orcids": []}, "institutions": [{"id": "https://openalex.org/I183067930", "display_name": "Shanghai Jiao Tong University", "ror": "https://ror.org/0220qvk04", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I183067930"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Shen, Wei", "raw_affiliation_strings": ["Shanghai Jiao Tong University ,"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Shanghai Jiao Tong University ,", "institution_ids": ["https://openalex.org/I183067930"]}]}], "abstract_inverted_index": {"The": [0], "rapid": [1], "development": [2], "of": [3, 110, 161], "large": [4], "language": [5, 10], "models": [6], "has": [7], "revolutionized": [8], "natural": [9], "processing,": [11], "but": [12], "their": [13, 81], "fine-tuning": [14, 22], "remains": [15], "computationally": [16], "expensive,": [17], "hindering": [18], "broad": [19], "deployment.": [20], "Parameter-efficient": [21], "(PEFT)": [23], "methods,": [24, 145], "such": [25], "as": [26, 30, 76, 168], "LoRA,": [27], "have": [28], "emerged": [29], "solutions.": [31], "Recent": [32], "work": [33], "like": [34], "DoRA": [35], "attempts": [36], "to": [37, 105, 152], "further": [38], "decompose": [39], "weight": [40, 74], "adaptation": [41, 82], "into": [42, 83, 129], "direction": [43, 52, 84], "and": [44, 79, 85, 100, 113, 121, 124, 159], "magnitude": [45, 86, 109], "components.": [46], "However,": [47], "existing": [48, 130, 144, 153], "formulations": [49], "often": [50], "define": [51], "heuristically": [53], "at": [54], "the": [55, 93, 108, 111, 157], "column": [56], "level,": [57], "lacking": [58], "a": [59, 69, 88, 97, 147, 169], "principled": [60], "geometric": [61], "foundation.": [62], "In": [63], "this": [64], "paper,": [65], "we": [66, 163], "propose": [67], "MAP,": [68, 162], "novel": [70], "framework": [71], "that": [72, 136], "reformulates": [73], "matrices": [75], "high-dimensional": [77], "vectors": [78], "decouples": [80], "in": [87], "rigorous": [89], "manner.": [90], "MAP": [91, 137], "normalizes": [92], "pre-trained": [94], "weights,": [95], "learns": [96], "directional": [98], "update,": [99], "introduces": [101], "two": [102], "scalar": [103], "coefficients": [104], "independently": [106], "scale": [107], "base": [112], "update": [114], "vectors.": [115], "This": [116], "design": [117], "enables": [118], "more": [119], "interpretable": [120], "flexible": [122], "adaptation,": [123], "can": [125, 166], "be": [126], "seamlessly": [127], "integrated": [128], "PEFT": [131, 154, 175], "methods.": [132, 155, 176], "Extensive": [133], "experiments": [134], "show": [135], "significantly": [138], "improves": [139], "performance": [140], "when": [141], "coupling": [142], "with": [143], "offering": [146], "simple": [148], "yet": [149], "powerful": [150], "enhancement": [151], "Given": [156], "universality": [158], "simplicity": [160], "hope": [164], "it": [165], "serve": [167], "default": [170], "setting": [171], "for": [172], "designing": [173], "future": [174]}, "primary_topic": {"id": "https://openalex.org/T11307", "display_name": "Domain Adaptation and Few-Shot Learning", "score": 0.8618000149726868, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T11307", "display_name": "Domain Adaptation and Few-Shot Learning", "score": 0.8618000149726868, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.12460000067949295, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10181", "display_name": "Natural Language Processing Techniques", "score": 0.005499999970197678, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "pmh:oai:arXiv.org:2505.23094", "is_oa": true, "landing_page_url": "http://arxiv.org/abs/2505.23094", "pdf_url": "https://arxiv.org/pdf/2505.23094", "source": {"id": "https://openalex.org/S4306400194", "display_name": "arXiv (Cornell University)", "issn_l": "2331-8422", "issn": ["2331-8422"], "is_oa": true, "is_in_doaj": false, "is_core": false, "listed_in": [], "host_organization": "https://openalex.org/I205783295", "host_organization_name": "Cornell University", "host_organization_lineage": ["https://openalex.org/I205783295"], "host_organization_lineage_names": [], "type": "repository"}, "license": null, "license_id": null, "version": "submittedVersion", "is_accepted": false, "is_published": false, "raw_source_name": null, "raw_type": "text"}, "cited_by_count": 0, "is_retracted": false}, {"id": "https://openalex.org/W4416117447", "doi": "https://doi.org/10.48550/arxiv.2504.04726", "title": "Hard Negative Sampling via Large Language Models for Recommendation", "publication_year": 2025, "publication_date": "2025-04-07", "type": "preprint", "language": "en", "authorships": [{"author_position": "first", "author": {"id": "https://openalex.org/A5103184486", "display_name": "Chu Zhao", "orcid": "https://orcid.org/0009-0001-8716-8206", "observed_orcids": ["https://orcid.org/0009-0001-8716-8206"]}, "institutions": [{"id": "https://openalex.org/I9224756", "display_name": "Northeastern University", "ror": "https://ror.org/03awzbc87", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I9224756"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Zhao, Chu", "raw_affiliation_strings": ["Northeastern University Shenyang , China"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Northeastern University Shenyang , China", "institution_ids": ["https://openalex.org/I9224756"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5086494876", "display_name": "Enneng Yang", "orcid": "https://orcid.org/0000-0001-5419-5286", "observed_orcids": ["https://orcid.org/0000-0001-5419-5286"]}, "institutions": [{"id": "https://openalex.org/I9224756", "display_name": "Northeastern University", "ror": "https://ror.org/03awzbc87", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I9224756"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Yang, Enneng", "raw_affiliation_strings": ["Northeastern University Shenyang , China"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Northeastern University Shenyang , China", "institution_ids": ["https://openalex.org/I9224756"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5083156764", "display_name": "Yuting Liu", "orcid": "https://orcid.org/0000-0002-0774-0007", "observed_orcids": ["https://orcid.org/0000-0002-0774-0007", "https://orcid.org/0009-0003-1357-9138"]}, "institutions": [{"id": "https://openalex.org/I9224756", "display_name": "Northeastern University", "ror": "https://ror.org/03awzbc87", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I9224756"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Liu, Yuting", "raw_affiliation_strings": ["Northeastern University Shenyang , China"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Northeastern University Shenyang , China", "institution_ids": ["https://openalex.org/I9224756"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5000544808", "display_name": "Jianzhe Zhao", "orcid": "https://orcid.org/0000-0003-4492-5075", "observed_orcids": ["https://orcid.org/0000-0003-4492-5075"]}, "institutions": [{"id": "https://openalex.org/I9224756", "display_name": "Northeastern University", "ror": "https://ror.org/03awzbc87", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I9224756"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Zhao, Jianzhe", "raw_affiliation_strings": ["Northeastern University Shenyang , China"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Northeastern University Shenyang , China", "institution_ids": ["https://openalex.org/I9224756"]}]}, {"author_position": "last", "author": {"id": "https://openalex.org/A5007061198", "display_name": "Guibing Guo", "orcid": "https://orcid.org/0000-0002-1709-5056", "observed_orcids": ["https://orcid.org/0000-0002-1709-5056"]}, "institutions": [{"id": "https://openalex.org/I9224756", "display_name": "Northeastern University", "ror": "https://ror.org/03awzbc87", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I9224756"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Guo, Guibing", "raw_affiliation_strings": ["Northeastern University Shenyang , China"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Northeastern University Shenyang , China", "institution_ids": ["https://openalex.org/I9224756"]}]}], "abstract_inverted_index": {"Hard": [0, 48], "negative": [1, 73, 89], "sampling": [2, 74], "improves": [3], "recommendation": [4], "performance": [5], "by": [6], "accelerating": [7], "convergence": [8], "and": [9, 57, 122, 139, 147], "sharpening": [10], "the": [11, 59, 77, 110, 116], "decision": [12], "boundary.": [13], "However,": [14], "most": [15], "existing": [16], "methods": [17, 33, 138], "rely": [18], "on": [19, 128], "heuristic": [20], "strategies,": [21], "selecting": [22], "negatives": [23], "from": [24], "a": [25, 70], "fixed": [26], "candidate": [27], "pool.": [28], "Lacking": [29], "semantic": [30, 41, 72, 78], "awareness,": [31], "these": [32], "often": [34], "misclassify": [35], "items": [36], "that": [37, 91, 133], "align": [38], "with": [39, 98], "users'": [40], "interests": [42], "as": [43], "negatives,": [44], "resulting": [45], "in": [46], "False": [47], "Negative": [49], "Samples": [50], "(FHNS).": [51], "Such": [52], "FHNS": [53], "inject": [54], "noisy": [55], "supervision": [56], "hinder": [58], "model's": [60], "optimal": [61], "performance.": [62], "To": [63], "address": [64], "this": [65], "challenge,": [66], "we": [67, 104], "propose": [68], "HNLMRec,": [69], "generative": [71], "framework.": [75], "Leveraging": [76], "reasoning": [79], "capabilities": [80], "of": [81], "Large": [82], "Language": [83], "Models": [84], "(LLMs),": [85], "HNLMRec": [86, 134], "directly": [87], "generates": [88], "samples": [90], "are": [92], "behaviorally": [93], "distinct": [94], "yet": [95], "semantically": [96], "relevant": [97], "respect": [99], "to": [100, 118], "user": [101], "preferences.": [102], "Furthermore,": [103], "integrate": [105], "collaborative": [106], "filtering": [107], "signals": [108], "into": [109], "LLM": [111], "via": [112], "supervised": [113], "fine-tuning,": [114], "guiding": [115], "model": [117], "synthesize": [119], "more": [120], "reliable": [121], "informative": [123], "hard": [124], "negatives.": [125], "Extensive": [126], "experiments": [127], "multiple": [129], "real-world": [130], "datasets": [131], "demonstrate": [132], "significantly": [135], "outperforms": [136], "traditional": [137], "LLM-enhanced": [140], "baselines,": [141], "while": [142], "effectively": [143], "mitigating": [144], "popularity": [145], "bias": [146], "data": [148], "sparsity,": [149], "thereby": [150], "improving": [151], "generalization.": [152]}, "primary_topic": {"id": "https://openalex.org/T10203", "display_name": "Recommender Systems and Techniques", "score": 1.0, "subfield": {"id": "https://openalex.org/subfields/1710", "display_name": "Information Systems"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T10203", "display_name": "Recommender Systems and Techniques", "score": 1.0, "subfield": {"id": "https://openalex.org/subfields/1710", "display_name": "Information Systems"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10286", "display_name": "Information Retrieval and Search Behavior", "score": 0.0, "subfield": {"id": "https://openalex.org/subfields/1710", "display_name": "Information Systems"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T12101", "display_name": "Advanced Bandit Algorithms Research", "score": 0.0, "subfield": {"id": "https://openalex.org/subfields/1803", "display_name": "Management Science and Operations Research"}, "field": {"id": "https://openalex.org/fields/18", "display_name": "Decision Sciences"}, "domain": {"id": "https://openalex.org/domains/2", "display_name": "Social Sciences"}}], "primary_location": {"id": "pmh:oai:arXiv.org:2504.04726", "is_oa": true, "landing_page_url": "http://arxiv.org/abs/2504.04726", "pdf_url": "https://arxiv.org/pdf/2504.04726", "source": {"id": "https://openalex.org/S4306400194", "display_name": "arXiv (Cornell University)", "issn_l": "2331-8422", "issn": ["2331-8422"], "is_oa": true, "is_in_doaj": false, "is_core": false, "listed_in": [], "host_organization": "https://openalex.org/I205783295", "host_organization_name": "Cornell University", "host_organization_lineage": ["https://openalex.org/I205783295"], "host_organization_lineage_names": [], "type": "repository"}, "license": "cc-by", "license_id": "https://openalex.org/licenses/cc-by", "version": "submittedVersion", "is_accepted": false, "is_published": false, "raw_source_name": null, "raw_type": "text"}, "cited_by_count": 0, "is_retracted": false}, {"id": "https://openalex.org/W7109053367", "doi": "https://doi.org/10.5281/zenodo.17829773", "title": "Reinforcement Learning from Language Feedback: Aligning Large Language Models with Complex Human Values", "publication_year": 2025, "publication_date": "2025-12-05", "type": "article", "language": null, "authorships": [{"author_position": "first", "author": {"id": null, "display_name": "Revista, Zen", "orcid": "https://orcid.org/0009-0007-6299-2008", "observed_orcids": ["https://orcid.org/0009-0007-6299-2008"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Revista, Zen", "raw_affiliation_strings": [], "raw_orcid": "https://orcid.org/0009-0007-6299-2008", "affiliations": []}, {"author_position": "last", "author": {"id": null, "display_name": "IA, 10", "orcid": "https://orcid.org/0009-0007-6299-2008", "observed_orcids": ["https://orcid.org/0009-0007-6299-2008"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "IA, 10", "raw_affiliation_strings": [], "raw_orcid": "https://orcid.org/0009-0007-6299-2008", "affiliations": []}], "abstract_inverted_index": {"Large": [0], "Language": [1, 55], "Models": [2], "(LLMs)": [3], "have": [4], "demonstrated": [5], "remarkable": [6], "capabilities": [7], "in": [8, 123], "generating": [9], "human-like": [10], "text,": [11], "but": [12], "aligning": [13, 70], "them": [14], "with": [15, 115], "complex": [16], "human": [17, 44, 116], "values": [18, 117], "remains": [19], "a": [20, 49, 64], "significant": [21], "challenge.": [22], "Traditional": [23], "Reinforcement": [24, 52], "Learning": [25, 53], "from": [26, 54], "Human": [27], "Feedback": [28, 56], "(RLHF)": [29], "relies": [30], "on": [31, 99], "scalar": [32], "rewards,": [33], "which": [34], "can": [35, 87, 110], "be": [36], "insufficient": [37], "to": [38, 112, 119, 145, 159], "capture": [39], "the": [40, 89, 96, 136], "nuances": [41], "of": [42, 91, 140], "multifaceted": [43], "preferences.": [45], "This": [46], "paper": [47], "explores": [48], "novel": [50], "approach:": [51], "(RLLF).": [57], "RLLF": [58, 109], "leverages": [59], "natural": [60], "language": [61, 77, 92], "feedback": [62, 78, 146], "as": [63], "richer,": [65], "more": [66], "expressive": [67], "signal": [68], "for": [69, 75], "LLMs.": [71], "We": [72, 133], "investigate": [73], "methods": [74], "encoding": [76], "into": [79], "reward": [80, 148], "functions,": [81], "designing": [82], "effective": [83], "training": [84], "algorithms": [85], "that": [86, 108], "handle": [88], "complexity": [90], "data,": [93], "and": [94, 130, 138, 150, 153], "evaluating": [95], "resulting": [97], "models": [98], "tasks": [100], "requiring": [101], "nuanced": [102], "value": [103], "judgments.": [104], "Our": [105], "experiments": [106], "demonstrate": [107], "lead": [111], "improved": [113], "alignment": [114], "compared": [118], "traditional": [120], "RLHF,": [121], "particularly": [122], "scenarios": [124], "involving": [125], "ethical": [126], "considerations,": [127], "subjective": [128], "preferences,": [129], "contextual": [131], "understanding.": [132], "also": [134], "analyze": [135], "challenges": [137], "limitations": [139], "RLLF,": [141], "including": [142], "issues": [143], "related": [144], "bias,": [147], "gaming,": [149], "computational": [151], "efficiency,": [152], "we": [154], "propose": [155], "future": [156], "research": [157], "directions": [158], "address": [160], "these": [161], "challenges.": [162]}, "primary_topic": {"id": "https://openalex.org/T10462", "display_name": "Reinforcement Learning in Robotics", "score": 0.5756999850273132, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T10462", "display_name": "Reinforcement Learning in Robotics", "score": 0.5756999850273132, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10883", "display_name": "Ethics and Social Impacts of AI", "score": 0.13609999418258667, "subfield": {"id": "https://openalex.org/subfields/3311", "display_name": "Safety Research"}, "field": {"id": "https://openalex.org/fields/33", "display_name": "Social Sciences"}, "domain": {"id": "https://openalex.org/domains/2", "display_name": "Social Sciences"}}, {"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.0681999996304512, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "doi:10.5281/zenodo.17829773", "is_oa": true, "landing_page_url": "https://doi.org/10.5281/zenodo.17829773", "pdf_url": null, "source": {"id": "https://openalex.org/S4306400562", "display_name": "Zenodo (CERN European Organization for Nuclear Research)", "issn_l": null, "issn": null, "is_oa": true, "is_in_doaj": false, "is_core": false, "listed_in": [], "host_organization": "https://openalex.org/I67311998", "host_organization_name": "European Organization for Nuclear Research", "host_organization_lineage": ["https://openalex.org/I67311998"], "host_organization_lineage_names": [], "type": "repository"}, "license": "cc-by", "license_id": "https://openalex.org/licenses/cc-by", "version": null, "is_accepted": false, "is_published": false, "raw_source_name": null, "raw_type": "JournalArticle"}, "cited_by_count": 0, "is_retracted": false}, {"id": "https://openalex.org/W4415962434", "doi": "https://doi.org/10.48550/arxiv.2510.17157", "title": "GACO-CAD: Geometry-Augmented and Conciseness-Optimized CAD Model Generation from Single Image", "publication_year": 2025, "publication_date": "2025-10-20", "type": "preprint", "language": null, "authorships": [{"author_position": "first", "author": {"id": "https://openalex.org/A5100443381", "display_name": "Yinghui Wang", "orcid": "https://orcid.org/0000-0001-6368-8568", "observed_orcids": ["https://orcid.org/0000-0001-6368-8568"]}, "institutions": [{"id": "https://openalex.org/I66867065", "display_name": "East China Normal University", "ror": "https://ror.org/02n96ep67", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I66867065"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Wang, Yinghui", "raw_affiliation_strings": ["East China Normal University , China"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "East China Normal University , China", "institution_ids": ["https://openalex.org/I66867065"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5013190724", "display_name": "Xinyu Zhang", "orcid": null, "observed_orcids": []}, "institutions": [{"id": "https://openalex.org/I66867065", "display_name": "East China Normal University", "ror": "https://ror.org/02n96ep67", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I66867065"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Zhang, Xinyu", "raw_affiliation_strings": ["East China Normal University , China"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "East China Normal University , China", "institution_ids": ["https://openalex.org/I66867065"]}]}, {"author_position": "last", "author": {"id": "https://openalex.org/A5101979071", "display_name": "Peng Du", "orcid": "https://orcid.org/0000-0001-6626-6627", "observed_orcids": ["https://orcid.org/0000-0001-6626-6627", "https://orcid.org/0000-0002-7658-233X"]}, "institutions": [{"id": "https://openalex.org/I76130692", "display_name": "Zhejiang University", "ror": "https://ror.org/00a2xv884", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I76130692"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Du, Peng", "raw_affiliation_strings": ["Zhejiang University , China"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Zhejiang University , China", "institution_ids": ["https://openalex.org/I76130692"]}]}], "abstract_inverted_index": {"Generating": [0], "editable,": [1], "parametric": [2, 157], "CAD": [3, 71], "models": [4, 25, 72], "from": [5, 34, 129], "a": [6, 50, 60, 105, 138], "single": [7], "image": [8, 102], "holds": [9], "great": [10], "potential": [11], "to": [12, 38, 58, 103, 167], "lower": [13], "the": [14, 65, 69, 75, 100, 109, 122, 149, 172, 184], "barriers": [15], "of": [16, 68, 77, 111, 151, 194], "industrial": [17], "concept": [18], "design.": [19], "However,": [20], "current": [21], "multi-modal": [22], "large": [23], "language": [24], "(MLLMs)": [26], "still": [27], "struggle": [28], "with": [29, 99], "accurately": [30], "inferring": [31], "3D": [32, 127], "geometry": [33, 128], "2D": [35, 130], "images": [36], "due": [37], "limited": [39], "spatial": [40, 118], "reasoning": [41], "capabilities.": [42], "We": [43], "address": [44], "this": [45], "limitation": [46], "by": [47], "introducing": [48], "GACO-CAD,": [49], "novel": [51], "two-stage": [52], "post-training": [53], "framework.": [54], "It": [55], "is": [56, 165], "designed": [57], "achieve": [59], "joint": [61], "objective:": [62], "simultaneously": [63], "improving": [64], "geometric": [66, 95, 146, 197], "accuracy": [67], "generated": [70], "and": [73, 89, 154, 174, 199], "encouraging": [74], "use": [76], "more": [78, 124, 152], "concise": [79], "modeling": [80, 158, 200], "procedures.": [81], "First,": [82], "during": [83, 133], "supervised": [84], "fine-tuning,": [85], "we": [86, 136], "leverage": [87], "depth": [88], "surface": [90], "normal": [91], "maps": [92], "as": [93], "dense": [94], "priors,": [96], "combining": [97], "them": [98], "RGB": [101], "form": [104], "multi-channel": [106], "input.": [107], "In": [108], "context": [110], "single-view": [112], "reconstruction,": [113], "these": [114], "priors": [115], "provide": [116], "complementary": [117], "cues": [119], "that": [120, 178], "help": [121], "MLLM": [123, 186], "reliably": [125], "recover": [126], "observations.": [131], "Second,": [132], "reinforcement": [134], "learning,": [135], "introduce": [137], "group": [139], "length": [140], "reward": [141], "that,": [142], "while": [143], "preserving": [144], "high": [145], "fidelity,": [147], "promotes": [148], "generation": [150], "compact": [153], "less": [155], "redundant": [156], "sequences.": [159], "A": [160], "simple": [161], "dynamic": [162], "weighting": [163], "strategy": [164], "adopted": [166], "stabilize": [168], "training.": [169], "Experiments": [170], "on": [171], "DeepCAD": [173], "Fusion360": [175], "datasets": [176], "show": [177], "GACO-CAD": [179], "achieves": [180], "state-of-the-art": [181], "performance": [182], "under": [183], "same": [185], "backbone,": [187], "consistently": [188], "outperforming": [189], "existing": [190], "methods": [191], "in": [192], "terms": [193], "code": [195], "validity,": [196], "accuracy,": [198], "conciseness.": [201]}, "primary_topic": {"id": "https://openalex.org/T10719", "display_name": "3D Shape Modeling and Analysis", "score": 0.9901999831199646, "subfield": {"id": "https://openalex.org/subfields/2206", "display_name": "Computational Mechanics"}, "field": {"id": "https://openalex.org/fields/22", "display_name": "Engineering"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T10719", "display_name": "3D Shape Modeling and Analysis", "score": 0.9901999831199646, "subfield": {"id": "https://openalex.org/subfields/2206", "display_name": "Computational Mechanics"}, "field": {"id": "https://openalex.org/fields/22", "display_name": "Engineering"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10775", "display_name": "Generative Adversarial Networks and Image Synthesis", "score": 0.003000000026077032, "subfield": {"id": "https://openalex.org/subfields/1707", "display_name": "Computer Vision and Pattern Recognition"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T11714", "display_name": "Multimodal Machine Learning Applications", "score": 0.0013000000035390258, "subfield": {"id": "https://openalex.org/subfields/1707", "display_name": "Computer Vision and Pattern Recognition"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "pmh:oai:arXiv.org:2510.17157", "is_oa": true, "landing_page_url": "http://arxiv.org/abs/2510.17157", "pdf_url": "https://arxiv.org/pdf/2510.17157", "source": {"id": "https://openalex.org/S4306400194", "display_name": "arXiv (Cornell University)", "issn_l": "2331-8422", "issn": ["2331-8422"], "is_oa": true, "is_in_doaj": false, "is_core": false, "listed_in": [], "host_organization": "https://openalex.org/I205783295", "host_organization_name": "Cornell University", "host_organization_lineage": ["https://openalex.org/I205783295"], "host_organization_lineage_names": [], "type": "repository"}, "license": "cc-by", "license_id": "https://openalex.org/licenses/cc-by", "version": "submittedVersion", "is_accepted": false, "is_published": false, "raw_source_name": null, "raw_type": "text"}, "cited_by_count": 0, "is_retracted": false}, {"id": "https://openalex.org/W7113915253", "doi": null, "title": "Investigating Training and Generalization in Faithful Self-Explanations of Large Language Models", "publication_year": 2025, "publication_date": "2025-12-08", "type": "preprint", "language": null, "authorships": [{"author_position": "first", "author": {"id": null, "display_name": "Doi, Tomoki", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Doi, Tomoki", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": null, "display_name": "Isonuma, Masaru", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Isonuma, Masaru", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "last", "author": {"id": null, "display_name": "Yanaka, Hitomi", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Yanaka, Hitomi", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}], "abstract_inverted_index": {"Large": [0], "language": [1], "models": [2], "have": [3, 58], "the": [4, 31, 44, 80, 88, 156], "potential": [5], "to": [6, 48, 90, 111, 155, 160, 177], "generate": [7], "explanations": [8, 107], "for": [9, 84, 124], "their": [10], "own": [11], "predictions": [12], "in": [13, 68, 181], "a": [14, 115, 178], "variety": [15], "of": [16, 46, 82, 153], "styles": [17, 57], "based": [18], "on": [19, 127], "user": [20], "instructions.": [21], "Recent": [22], "research": [23], "has": [24, 36], "examined": [25], "whether": [26, 65], "these": [27, 92, 121, 148], "self-explanations": [28, 86, 123], "faithfully": [29], "reflect": [30], "models'": [32], "actual": [33], "behavior": [34], "and": [35, 87, 99, 119, 143, 146, 159], "found": [37], "that": [38, 108, 133, 147, 173], "they": [39], "often": [40], "lack": [41], "faithfulness.": [42], "However,": [43], "question": [45], "how": [47], "improve": [49, 136], "faithfulness": [50, 138], "remains": [51], "underexplored.": [52], "Moreover,": [53], "because": [54], "different": [55], "explanation": [56, 101, 144], "superficially": [59], "distinct": [60], "characteristics,": [61], "it": [62], "is": [63], "unclear": [64], "improvements": [66, 149], "observed": [67], "one": [69], "style": [70], "also": [71, 150], "arise": [72], "when": [73], "using": [74, 95, 114], "other": [75], "styles.": [76, 102], "This": [77], "study": [78], "analyzes": [79], "effects": [81, 93], "training": [83, 134, 174], "faithful": [85, 113, 182], "extent": [89], "which": [91], "generalize,": [94], "three": [96, 100, 170], "classification": [97, 141], "tasks": [98, 142], "We": [103], "construct": [104], "one-word": [105], "constrained": [106], "are": [109], "likely": [110], "be": [112], "feature": [116], "attribution": [117], "method,": [118], "use": [120], "pseudo-faithful": [122], "continual": [125], "learning": [126], "instruction-tuned": [128], "models.": [129], "Our": [130], "experiments": [131], "demonstrate": [132], "can": [135], "self-explanation": [137, 183], "across": [139], "all": [140], "styles,": [145, 171], "show": [151], "signs": [152], "generalization": [154, 168], "multi-word": [157], "settings": [158], "unseen": [161], "tasks.": [162], "Furthermore,": [163], "we": [164], "find": [165], "consistent": [166], "cross-style": [167], "among": [169], "suggesting": [172], "may": [175], "contribute": [176], "broader": [179], "improvement": [180], "ability.": [184]}, "primary_topic": {"id": "https://openalex.org/T12026", "display_name": "Explainable Artificial Intelligence (XAI)", "score": 1.0, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T12026", "display_name": "Explainable Artificial Intelligence (XAI)", "score": 1.0, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 9.999999747378752e-05, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T11689", "display_name": "Adversarial Robustness in Machine Learning", "score": 0.0, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "pmh:oai:arXiv.org:2512.07288", "is_oa": true, "landing_page_url": "http://arxiv.org/abs/2512.07288", "pdf_url": "https://arxiv.org/pdf/2512.07288", "source": {"id": "https://openalex.org/S4306400194", "display_name": "arXiv (Cornell University)", "issn_l": "2331-8422", "issn": ["2331-8422"], "is_oa": true, "is_in_doaj": false, "is_core": false, "listed_in": [], "host_organization": "https://openalex.org/I205783295", "host_organization_name": "Cornell University", "host_organization_lineage": ["https://openalex.org/I205783295"], "host_organization_lineage_names": [], "type": "repository"}, "license": null, "license_id": null, "version": "submittedVersion", "is_accepted": false, "is_published": false, "raw_source_name": null, "raw_type": "text"}, "cited_by_count": 0, "is_retracted": false}, {"id": "https://openalex.org/W4415818364", "doi": "https://doi.org/10.3390/app152111732", "title": "From Knowledge Extraction to Assertive Response: An LLM Chatbot for Information Retrieval in Telemedicine Systems", "publication_year": 2025, "publication_date": "2025-11-03", "type": "article", "language": "en", "authorships": [{"author_position": "first", "author": {"id": "https://openalex.org/A5120234774", "display_name": "Bruna D. Pupo", "orcid": "https://orcid.org/0009-0006-4977-4463", "observed_orcids": ["https://orcid.org/0009-0006-4977-4463"]}, "institutions": [{"id": "https://openalex.org/I4104125", "display_name": "Universidade Federal de Santa Catarina", "ror": "https://ror.org/041akq887", "country_code": "BR", "type": "education", "lineage": ["https://openalex.org/I4104125"]}], "countries": ["BR"], "is_corresponding": false, "raw_author_name": "Bruna D. Pupo", "raw_affiliation_strings": ["Telemedicine Laboratory, Federal University of Santa Catarina (UFSC), Florian\u00f3polis 88040-900, Brazil", "Brazilian Institute for Digital Convergence (INCoD), Florian\u00f3polis 88040-900, Brazil"], "raw_orcid": "https://orcid.org/0009-0006-4977-4463", "affiliations": [{"raw_affiliation_string": "Telemedicine Laboratory, Federal University of Santa Catarina (UFSC), Florian\u00f3polis 88040-900, Brazil", "institution_ids": ["https://openalex.org/I4104125"]}, {"raw_affiliation_string": "Brazilian Institute for Digital Convergence (INCoD), Florian\u00f3polis 88040-900, Brazil", "institution_ids": []}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5068969835", "display_name": "Daniel Gouveia Costa", "orcid": "https://orcid.org/0000-0003-3988-8476", "observed_orcids": ["https://orcid.org/0000-0003-3988-8476"]}, "institutions": [{"id": "https://openalex.org/I182534213", "display_name": "Universidade do Porto", "ror": "https://ror.org/043pwc612", "country_code": "PT", "type": "education", "lineage": ["https://openalex.org/I182534213"]}, {"id": "https://openalex.org/I4405252769", "display_name": "Centro de Investiga\u00e7\u00e3o em Sistemas e Tecnologias", "ror": "https://ror.org/00tpr5718", "country_code": null, "type": "education", "lineage": ["https://openalex.org/I4405252769"]}], "countries": ["PT"], "is_corresponding": true, "raw_author_name": "Daniel G. Costa", "raw_affiliation_strings": ["SYSTEC-ARISE, Faculty of Engineering, University of Porto, 4200-465 Porto, Portugal"], "raw_orcid": "https://orcid.org/0000-0003-3988-8476", "affiliations": [{"raw_affiliation_string": "SYSTEC-ARISE, Faculty of Engineering, University of Porto, 4200-465 Porto, Portugal", "institution_ids": ["https://openalex.org/I182534213", "https://openalex.org/I4405252769"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5023566455", "display_name": "Roger Immich", "orcid": "https://orcid.org/0000-0003-2483-6382", "observed_orcids": ["https://orcid.org/0000-0003-2483-6382"]}, "institutions": [{"id": "https://openalex.org/I35046152", "display_name": "Universidade Federal do Rio Grande do Norte", "ror": "https://ror.org/04wn09761", "country_code": "BR", "type": "education", "lineage": ["https://openalex.org/I35046152"]}], "countries": ["BR"], "is_corresponding": false, "raw_author_name": "Roger Immich", "raw_affiliation_strings": ["Instituto Metr\u00f3pole Digital (IMD/UFRN), Natal 59078-900, Brazil"], "raw_orcid": "https://orcid.org/0000-0003-2483-6382", "affiliations": [{"raw_affiliation_string": "Instituto Metr\u00f3pole Digital (IMD/UFRN), Natal 59078-900, Brazil", "institution_ids": ["https://openalex.org/I35046152"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5004153371", "display_name": "Aldo von Wangenheim", "orcid": "https://orcid.org/0000-0003-4532-1417", "observed_orcids": ["https://orcid.org/0000-0003-4532-1417"]}, "institutions": [{"id": "https://openalex.org/I4104125", "display_name": "Universidade Federal de Santa Catarina", "ror": "https://ror.org/041akq887", "country_code": "BR", "type": "education", "lineage": ["https://openalex.org/I4104125"]}], "countries": ["BR"], "is_corresponding": false, "raw_author_name": "Aldo von Wangenheim", "raw_affiliation_strings": ["Brazilian Institute for Digital Convergence (INCoD), Florian\u00f3polis 88040-900, Brazil", "Telemedicine Laboratory, Federal University of Santa Catarina (UFSC), Florian\u00f3polis 88040-900, Brazil"], "raw_orcid": "https://orcid.org/0000-0003-4532-1417", "affiliations": [{"raw_affiliation_string": "Brazilian Institute for Digital Convergence (INCoD), Florian\u00f3polis 88040-900, Brazil", "institution_ids": []}, {"raw_affiliation_string": "Telemedicine Laboratory, Federal University of Santa Catarina (UFSC), Florian\u00f3polis 88040-900, Brazil", "institution_ids": ["https://openalex.org/I4104125"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5067873815", "display_name": "Alex Sandro Roschildt Pinto", "orcid": "https://orcid.org/0000-0002-9144-1535", "observed_orcids": ["https://orcid.org/0000-0002-9144-1535"]}, "institutions": [{"id": "https://openalex.org/I4104125", "display_name": "Universidade Federal de Santa Catarina", "ror": "https://ror.org/041akq887", "country_code": "BR", "type": "education", "lineage": ["https://openalex.org/I4104125"]}], "countries": ["BR"], "is_corresponding": false, "raw_author_name": "Alex Sandro Roschildt Pinto", "raw_affiliation_strings": ["Brazilian Institute for Digital Convergence (INCoD), Florian\u00f3polis 88040-900, Brazil", "Telemedicine Laboratory, Federal University of Santa Catarina (UFSC), Florian\u00f3polis 88040-900, Brazil"], "raw_orcid": "https://orcid.org/0000-0002-9144-1535", "affiliations": [{"raw_affiliation_string": "Brazilian Institute for Digital Convergence (INCoD), Florian\u00f3polis 88040-900, Brazil", "institution_ids": []}, {"raw_affiliation_string": "Telemedicine Laboratory, Federal University of Santa Catarina (UFSC), Florian\u00f3polis 88040-900, Brazil", "institution_ids": ["https://openalex.org/I4104125"]}]}, {"author_position": "last", "author": {"id": "https://openalex.org/A5080870100", "display_name": "Douglas Dyllon Jer\u00f4nimo de Macedo", "orcid": "https://orcid.org/0000-0002-3237-4168", "observed_orcids": ["https://orcid.org/0000-0002-3237-4168"]}, "institutions": [{"id": "https://openalex.org/I4104125", "display_name": "Universidade Federal de Santa Catarina", "ror": "https://ror.org/041akq887", "country_code": "BR", "type": "education", "lineage": ["https://openalex.org/I4104125"]}], "countries": ["BR"], "is_corresponding": false, "raw_author_name": "Douglas D. J. de Macedo", "raw_affiliation_strings": ["Brazilian Institute for Digital Convergence (INCoD), Florian\u00f3polis 88040-900, Brazil", "Telemedicine Laboratory, Federal University of Santa Catarina (UFSC), Florian\u00f3polis 88040-900, Brazil"], "raw_orcid": "https://orcid.org/0000-0002-3237-4168", "affiliations": [{"raw_affiliation_string": "Brazilian Institute for Digital Convergence (INCoD), Florian\u00f3polis 88040-900, Brazil", "institution_ids": []}, {"raw_affiliation_string": "Telemedicine Laboratory, Federal University of Santa Catarina (UFSC), Florian\u00f3polis 88040-900, Brazil", "institution_ids": ["https://openalex.org/I4104125"]}]}], "abstract_inverted_index": {"The": [0, 69], "development": [1], "of": [2, 15, 19, 104, 160, 173], "new": [3, 17, 32], "technologies,": [4], "improved": [5], "by": [6], "advances": [7], "in": [8, 21, 176, 183], "artificial": [9], "intelligence,": [10], "has": [11, 30], "enabled": [12, 155], "the": [13, 59, 102, 167, 171], "creation": [14], "a": [16, 45, 81, 132, 147, 156, 161], "generation": [18, 134], "applications": [20], "different": [22], "scenarios.": [23], "In": [24, 42, 110], "medical": [25], "systems,": [26, 178], "adopting": [27], "AI-driven": [28], "solutions": [29], "brought": [31], "possibilities,": [33], "but": [34], "their": [35], "effective": [36], "impacts": [37], "still": [38], "need": [39], "further": [40], "investigation.": [41], "this": [43], "context,": [44], "chatbot": [46], "prototype": [47], "trained": [48], "with": [49, 78], "large": [50], "language": [51], "models": [52], "(LLMs)": [53], "was": [54], "developed": [55], "using": [56], "data": [57], "from": [58], "Santa": [60], "Catarina": [61], "Telemedicine": [62], "and": [63, 93, 123, 170], "Telehealth": [64], "System": [65], "(STT)": [66], "Dermatology": [67], "module.": [68], "system": [70], "adapts": [71], "Llama": [72], "3": [73], "8B": [74], "via": [75], "supervised": [76], "Fine-tuning": [77], "QLoRA": [79], "on": [80, 107], "proprietary,": [82], "domain-specific": [83], "dataset": [84], "(33": [85], "input-output": [86], "pairs).": [87], "Although": [88], "it": [89], "achieved": [90], "100%": [91], "Fluency": [92], "89.74%": [94], "Coherence,": [95], "Factual": [96], "Correctness": [97], "remained": [98], "low": [99, 127], "(43.59%),": [100], "highlighting": [101, 165], "limitations": [103], "training": [105], "LLMs": [106, 175], "small": [108], "datasets.": [109], "addition": [111], "to": [112], "G-Eval": [113], "metrics,": [114], "we": [115, 144], "conducted": [116], "expert": [117], "human": [118], "validation,": [119], "encompassing": [120], "both": [121, 166], "quantitative": [122], "qualitative": [124], "aspects.": [125], "This": [126, 153], "factual": [128], "score": [129], "indicates": [130], "that": [131], "retrieval-augmented": [133], "(RAG)": [135], "mechanism": [136], "is": [137], "essential": [138], "for": [139, 150], "robust": [140], "information": [141], "retrieval,": [142], "which": [143], "outline": [145], "as": [146, 180], "primary": [148], "direction": [149], "future": [151], "work.": [152], "approach": [154], "more": [157], "in-depth": [158], "analysis": [159], "real-world": [162], "telemedicine": [163], "environment,": [164], "practical": [168], "challenges": [169], "benefits": [172], "implementing": [174], "complex": [177], "such": [179], "those": [181], "used": [182], "telemedicine.": [184]}, "primary_topic": {"id": "https://openalex.org/T11636", "display_name": "Artificial Intelligence in Healthcare and Education", "score": 0.9824000000953674, "subfield": {"id": "https://openalex.org/subfields/2718", "display_name": "Health Informatics"}, "field": {"id": "https://openalex.org/fields/27", "display_name": "Medicine"}, "domain": {"id": "https://openalex.org/domains/4", "display_name": "Health Sciences"}}, "topics": [{"id": "https://openalex.org/T11636", "display_name": "Artificial Intelligence in Healthcare and Education", "score": 0.9824000000953674, "subfield": {"id": "https://openalex.org/subfields/2718", "display_name": "Health Informatics"}, "field": {"id": "https://openalex.org/fields/27", "display_name": "Medicine"}, "domain": {"id": "https://openalex.org/domains/4", "display_name": "Health Sciences"}}, {"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.004800000227987766, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10912", "display_name": "Telemedicine and Telehealth Implementation", "score": 0.002300000051036477, "subfield": {"id": "https://openalex.org/subfields/2739", "display_name": "Public Health, Environmental and Occupational Health"}, "field": {"id": "https://openalex.org/fields/27", "display_name": "Medicine"}, "domain": {"id": "https://openalex.org/domains/4", "display_name": "Health Sciences"}}], "primary_location": {"id": "doi:10.3390/app152111732", "is_oa": true, "landing_page_url": "https://doi.org/10.3390/app152111732", "pdf_url": "https://www.mdpi.com/2076-3417/15/21/11732/pdf?version=1762181231", "source": {"id": "https://openalex.org/S4210205812", "display_name": "Applied Sciences", "issn_l": "2076-3417", "issn": ["2076-3417"], "is_oa": true, "is_in_doaj": true, "is_core": true, "listed_in": ["cwts-core", "doaj", "norway-1"], "host_organization": "https://openalex.org/P4310310987", "host_organization_name": "Multidisciplinary Digital Publishing Institute", "host_organization_lineage": ["https://openalex.org/P4310310987"], "host_organization_lineage_names": ["Multidisciplinary Digital Publishing Institute"], "type": "journal"}, "license": "cc-by", "license_id": "https://openalex.org/licenses/cc-by", "version": "publishedVersion", "is_accepted": true, "is_published": true, "raw_source_name": "Applied Sciences", "raw_type": "journal-article"}, "cited_by_count": 2, "is_retracted": false}, {"id": "https://openalex.org/W4407632334", "doi": "https://doi.org/10.48550/arxiv.2502.09838", "title": "HealthGPT: A Medical Large Vision-Language Model for Unifying Comprehension and Generation via Heterogeneous Knowledge Adaptation", "publication_year": 2025, "publication_date": "2025-02-14", "type": "preprint", "language": "en", "authorships": [{"author_position": "first", "author": {"id": null, "display_name": "Lin, Tianwei", "orcid": null, "observed_orcids": []}, "institutions": [{"id": "https://openalex.org/I76130692", "display_name": "Zhejiang University", "ror": "https://ror.org/00a2xv884", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I76130692"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Lin, Tianwei", "raw_affiliation_strings": ["Zhejiang University ,"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Zhejiang University ,", "institution_ids": ["https://openalex.org/I76130692"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5082131901", "display_name": "Wenqiao Zhang", "orcid": "https://orcid.org/0000-0002-5988-7609", "observed_orcids": ["https://orcid.org/0000-0002-5988-7609"]}, "institutions": [{"id": "https://openalex.org/I76130692", "display_name": "Zhejiang University", "ror": "https://ror.org/00a2xv884", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I76130692"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Zhang, Wenqiao", "raw_affiliation_strings": ["Zhejiang University ,"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Zhejiang University ,", "institution_ids": ["https://openalex.org/I76130692"]}]}, {"author_position": "middle", "author": {"id": null, "display_name": "Li, Sijing", "orcid": null, "observed_orcids": []}, "institutions": [{"id": "https://openalex.org/I76130692", "display_name": "Zhejiang University", "ror": "https://ror.org/00a2xv884", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I76130692"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Li, Sijing", "raw_affiliation_strings": ["Zhejiang University ,"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Zhejiang University ,", "institution_ids": ["https://openalex.org/I76130692"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5121626354", "display_name": "Yuqian Yuan", "orcid": null, "observed_orcids": []}, "institutions": [{"id": "https://openalex.org/I76130692", "display_name": "Zhejiang University", "ror": "https://ror.org/00a2xv884", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I76130692"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Yuan, Yuqian", "raw_affiliation_strings": ["Zhejiang University ,"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Zhejiang University ,", "institution_ids": ["https://openalex.org/I76130692"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5088072182", "display_name": "Bo Yu", "orcid": "https://orcid.org/0000-0003-4689-9341", "observed_orcids": ["https://orcid.org/0000-0003-4689-9341", "https://orcid.org/0000-0001-6576-5555"]}, "institutions": [{"id": "https://openalex.org/I150229711", "display_name": "University of Electronic Science and Technology of China", "ror": "https://ror.org/04qr3zq92", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I150229711"]}, {"id": "https://openalex.org/I45928872", "display_name": "Alibaba Group (China)", "ror": "https://ror.org/00k642b80", "country_code": "CN", "type": "company", "lineage": ["https://openalex.org/I45928872"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Yu, Binhe", "raw_affiliation_strings": ["University of Electronic Science and Technology of China , 3 Alibaba ,"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "University of Electronic Science and Technology of China , 3 Alibaba ,", "institution_ids": ["https://openalex.org/I150229711", "https://openalex.org/I45928872"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5100673440", "display_name": "Haoyuan Li", "orcid": "https://orcid.org/0009-0007-4678-9923", "observed_orcids": ["https://orcid.org/0009-0007-4678-9923"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Li, Haoyuan", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5008173176", "display_name": "Wanggui He", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "He, Wanggui", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5115589203", "display_name": "Hao Jiang", "orcid": "https://orcid.org/0009-0009-9296-6067", "observed_orcids": ["https://orcid.org/0009-0009-9296-6067"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Jiang, Hao", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5101737334", "display_name": "Mengze Li", "orcid": "https://orcid.org/0000-0002-3482-234X", "observed_orcids": ["https://orcid.org/0000-0002-3482-234X"]}, "institutions": [{"id": "https://openalex.org/I200769079", "display_name": "Hong Kong University of Science and Technology", "ror": "https://ror.org/00q4vv597", "country_code": "HK", "type": "education", "lineage": ["https://openalex.org/I200769079"]}], "countries": ["HK"], "is_corresponding": false, "raw_author_name": "Li, Mengze", "raw_affiliation_strings": ["The Hong Kong University of Science and Technology ,"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "The Hong Kong University of Science and Technology ,", "institution_ids": ["https://openalex.org/I200769079"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5101906943", "display_name": "Xiaohui Song", "orcid": "https://orcid.org/0000-0001-7713-1373", "observed_orcids": ["https://orcid.org/0000-0001-7713-1373"]}, "institutions": [{"id": "https://openalex.org/I76130692", "display_name": "Zhejiang University", "ror": "https://ror.org/00a2xv884", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I76130692"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Song, Xiaohui", "raw_affiliation_strings": ["Zhejiang University ,"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Zhejiang University ,", "institution_ids": ["https://openalex.org/I76130692"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5063062444", "display_name": "Siliang Tang", "orcid": "https://orcid.org/0000-0002-7356-9711", "observed_orcids": ["https://orcid.org/0000-0002-7356-9711"]}, "institutions": [{"id": "https://openalex.org/I76130692", "display_name": "Zhejiang University", "ror": "https://ror.org/00a2xv884", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I76130692"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Tang, Siliang", "raw_affiliation_strings": ["Zhejiang University ,"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Zhejiang University ,", "institution_ids": ["https://openalex.org/I76130692"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5101485989", "display_name": "Jun Xiao", "orcid": "https://orcid.org/0000-0002-6142-9914", "observed_orcids": ["https://orcid.org/0000-0002-6142-9914"]}, "institutions": [{"id": "https://openalex.org/I76130692", "display_name": "Zhejiang University", "ror": "https://ror.org/00a2xv884", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I76130692"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Xiao, Jun", "raw_affiliation_strings": ["Zhejiang University ,"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Zhejiang University ,", "institution_ids": ["https://openalex.org/I76130692"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5101267417", "display_name": "Hui Lin", "orcid": "https://orcid.org/0009-0004-7343-2522", "observed_orcids": ["https://orcid.org/0009-0004-7343-2522"]}, "institutions": [{"id": "https://openalex.org/I76130692", "display_name": "Zhejiang University", "ror": "https://ror.org/00a2xv884", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I76130692"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Lin, Hui", "raw_affiliation_strings": ["Zhejiang University ,"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Zhejiang University ,", "institution_ids": ["https://openalex.org/I76130692"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5008666077", "display_name": "Yueting Zhuang", "orcid": "https://orcid.org/0000-0001-9017-2508", "observed_orcids": ["https://orcid.org/0000-0001-9017-2508"]}, "institutions": [{"id": "https://openalex.org/I76130692", "display_name": "Zhejiang University", "ror": "https://ror.org/00a2xv884", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I76130692"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Zhuang, Yueting", "raw_affiliation_strings": ["Zhejiang University ,"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Zhejiang University ,", "institution_ids": ["https://openalex.org/I76130692"]}]}, {"author_position": "last", "author": {"id": "https://openalex.org/A5024892041", "display_name": "Beng Chin Ooi", "orcid": "https://orcid.org/0000-0003-4446-1100", "observed_orcids": ["https://orcid.org/0000-0003-4446-1100"]}, "institutions": [{"id": "https://openalex.org/I165932596", "display_name": "National University of Singapore", "ror": "https://ror.org/01tgyzw49", "country_code": "SG", "type": "education", "lineage": ["https://openalex.org/I165932596"]}], "countries": ["SG"], "is_corresponding": false, "raw_author_name": "Ooi, Beng Chin", "raw_affiliation_strings": ["National University of Singapore Project"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "National University of Singapore Project", "institution_ids": ["https://openalex.org/I165932596"]}]}], "abstract_inverted_index": {"We": [0], "present": [1], "HealthGPT,": [2, 71], "a": [3, 19, 45, 56, 63, 74], "powerful": [4], "Medical": [5], "Large": [6], "Vision-Language": [7], "Model": [8], "(Med-LVLM)": [9], "that": [10], "integrates": [11], "medical": [12, 76, 94], "visual": [13, 59, 95], "comprehension": [14, 31, 78], "and": [15, 32, 62, 79, 89], "generation": [16, 33, 80], "capabilities": [17], "within": [18], "unified": [20, 96], "autoregressive": [21], "paradigm.": [22], "Our": [23, 98], "bootstrapping": [24], "philosophy": [25], "is": [26, 42, 53], "to": [27, 35], "progressively": [28], "adapt": [29], "heterogeneous": [30, 47], "knowledge": [34], "pre-trained": [36], "large": [37], "language": [38], "models": [39], "(LLMs).": [40], "This": [41], "achieved": [43], "through": [44], "novel": [46], "low-rank": [48], "adaptation": [49], "(H-LoRA)": [50], "technique,": [51], "which": [52], "complemented": [54], "by": [55], "tailored": [57], "hierarchical": [58], "perception": [60], "approach": [61], "three-stage": [64], "learning": [65], "strategy.": [66], "To": [67], "effectively": [68], "learn": [69], "the": [70], "we": [72], "devise": [73], "comprehensive": [75], "domain-specific": [77], "dataset": [81], "called": [82], "VL-Health.": [83], "Experimental": [84], "results": [85], "demonstrate": [86], "exceptional": [87], "performance": [88], "scalability": [90], "of": [91], "HealthGPT": [92], "in": [93], "tasks.": [97], "project": [99], "can": [100], "be": [101], "accessed": [102], "at": [103], "https://github.com/DCDmllm/HealthGPT.": [104]}, "primary_topic": {"id": "https://openalex.org/T11714", "display_name": "Multimodal Machine Learning Applications", "score": 0.9994999766349792, "subfield": {"id": "https://openalex.org/subfields/1707", "display_name": "Computer Vision and Pattern Recognition"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T11714", "display_name": "Multimodal Machine Learning Applications", "score": 0.9994999766349792, "subfield": {"id": "https://openalex.org/subfields/1707", "display_name": "Computer Vision and Pattern Recognition"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T11636", "display_name": "Artificial Intelligence in Healthcare and Education", "score": 0.0003000000142492354, "subfield": {"id": "https://openalex.org/subfields/2718", "display_name": "Health Informatics"}, "field": {"id": "https://openalex.org/fields/27", "display_name": "Medicine"}, "domain": {"id": "https://openalex.org/domains/4", "display_name": "Health Sciences"}}, {"id": "https://openalex.org/T11307", "display_name": "Domain Adaptation and Few-Shot Learning", "score": 0.0, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "pmh:oai:arXiv.org:2502.09838", "is_oa": true, "landing_page_url": "http://arxiv.org/abs/2502.09838", "pdf_url": "https://arxiv.org/pdf/2502.09838", "source": {"id": "https://openalex.org/S4306400194", "display_name": "arXiv (Cornell University)", "issn_l": "2331-8422", "issn": ["2331-8422"], "is_oa": true, "is_in_doaj": false, "is_core": false, "listed_in": [], "host_organization": "https://openalex.org/I205783295", "host_organization_name": "Cornell University", "host_organization_lineage": ["https://openalex.org/I205783295"], "host_organization_lineage_names": [], "type": "repository"}, "license": null, "license_id": null, "version": "submittedVersion", "is_accepted": false, "is_published": false, "raw_source_name": null, "raw_type": "text"}, "cited_by_count": 8, "is_retracted": false}, {"id": "https://openalex.org/W4412019468", "doi": "https://doi.org/10.2196/preprints.79039", "title": "Large Language Model\u2013Based Virtual Patient Systems for History-Taking in Medical Education: Comprehensive Systematic Review (Preprint)", "publication_year": 2025, "publication_date": "2025-06-14", "type": "preprint", "language": "en", "authorships": [{"author_position": "first", "author": {"id": "https://openalex.org/A5073380371", "display_name": "Dongliang Li", "orcid": "https://orcid.org/0009-0001-8675-3702", "observed_orcids": ["https://orcid.org/0009-0001-8675-3702"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Dongliang Li", "raw_affiliation_strings": [], "raw_orcid": "https://orcid.org/0009-0001-8675-3702", "affiliations": []}, {"author_position": "last", "author": {"id": "https://openalex.org/A5053859600", "display_name": "Syaheerah Lebai Lutfi", "orcid": "https://orcid.org/0000-0001-7349-0061", "observed_orcids": ["https://orcid.org/0000-0001-7349-0061"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Syaheerah Lebai Lutfi", "raw_affiliation_strings": [], "raw_orcid": "https://orcid.org/0000-0001-7349-0061", "affiliations": []}], "abstract_inverted_index": {"BACKGROUND": [0], "Large": [1], "language": [2], "models": [3], "(LLMs),": [4], "such": [5, 231], "as": [6, 232], "GPT-3.5": [7], "and": [8, 23, 68, 77, 80, 84, 95, 111, 118, 147, 180, 184, 191, 195, 202, 224, 246, 251, 266, 289, 294, 313, 332, 343], "GPT-4": [9], "(OpenAI),": [10], "have": [11], "been": [12], "transforming": [13], "virtual": [14, 54, 140, 271], "patient": [15, 55, 66, 141, 272, 283], "systems": [16, 56, 142, 273], "in": [17, 275], "medical": [18, 58], "education": [19], "by": [20, 132], "providing": [21], "scalable": [22], "cost-effective": [24], "alternatives": [25], "to": [26, 51, 105, 336], "standardized": [27, 328], "patients.": [28], "However,": [29, 219], "systematic": [30, 48], "evaluations": [31], "of": [32, 126, 260, 317], "their": [33], "performance,": [34], "particularly": [35], "for": [36, 57, 92, 237], "multimorbidity": [37, 161, 282], "scenarios": [38], "involving": [39, 199], "multiple": [40, 277], "coexisting": [41], "diseases,": [42], "are": [43], "still": [44], "limited.": [45], "OBJECTIVE": [46], "This": [47], "review": [49], "aimed": [50], "evaluate": [52], "LLM-based": [53, 139, 270], "history-taking,": [59], "addressing": [60, 153], "four": [61], "research": [62, 297], "questions:": [63], "(1)": [64], "simulated": [65, 144], "types": [67, 157, 279], "disease": [69, 156, 278, 292], "scope,": [70], "(2)": [71], "performance-enhancing": [72], "techniques,": [73], "(3)": [74], "experimental": [75], "designs": [76], "evaluation": [78], "metrics,": [79, 223, 263, 329], "(4)": [81], "dataset": [82, 268], "characteristics": [83], "availability.": [85], "METHODS": [86], "Following": [87], "PRISMA": [88], "(Preferred": [89], "Reporting": [90], "Items": [91], "Systematic": [93], "Reviews": [94], "Meta-Analyses)": [96], "2020,": [97, 104], "9": [98], "databases": [99], "were": [100, 114, 121, 129], "searched": [101], "(January": [102], "1,": [103], "August": [106], "18,": [107], "2025).": [108], "Nontransformer": [109], "LLMs": [110], "non\u2013history-taking": [112], "tasks": [113], "excluded.": [115], "Multidimensional": [116], "quality": [117], "bias": [119, 245], "assessments": [120], "conducted.": [122], "RESULTS": [123], "A": [124], "total": [125], "39": [127], "studies": [128, 256], "included,": [130], "screened": [131], "one": [133], "computer": [134], "science": [135], "researcher": [136], "under": [137], "supervision.": [138], "mainly": [143], "internal": [145], "medicine": [146], "mental": [148], "health": [149], "disorders,": [150], "with": [151, 304], "many": [152], "distinct": [154], "single": [155], "but": [158, 280], "few": [159], "covering": [160], "or": [162], "rare": [163], "conditions.": [164], "Techniques": [165], "like": [166], "role-based": [167], "prompts,": [168], "few-shot": [169], "learning,": [170], "multiagent": [171], "frameworks,": [172], "knowledge": [173], "graph": [174], "(KG)": [175], "integration": [176], "(top-k": [177, 208], "accuracy": [178, 288], "16.02%),": [179], "fine-tuning": [181], "enhanced": [182], "dialogue": [183, 324], "diagnostic": [185, 340], "accuracy.": [186], "Multimodal": [187], "inputs": [188], "(eg,": [189, 307], "speech": [190], "imaging)": [192], "improved": [193], "immersion": [194], "realism.": [196], "Evaluations,": [197], "typically": [198], "10-50": [200], "students": [201], "3-10": [203], "experts,": [204], "demonstrated": [205], "strong": [206], "performance": [207], "accuracy:": [209], "0.45-0.98,": [210], "hallucination": [211], "rate:": [212], "0.31%\u20135%,": [213], "System": [214], "Usability": [215], "Scale": [216], "[SUS]": [217], "\u226580).": [218], "small": [220, 264], "samples,": [221], "inconsistent": [222, 262], "limited": [225, 267], "controls": [226], "restricted": [227], "generalizability.": [228], "Common": [229], "datasets": [230, 335], "MIMIC-III": [233], "(Medical": [234], "Information": [235], "Mart": [236], "Intensive": [238], "Care-III)": [239], "exhibited": [240], "intensive": [241], "care": [242], "unit": [243], "(ICU)": [244], "lacked": [247], "diversity,": [248], "affecting": [249], "reproducibility": [250], "external": [252], "validity.": [253], "CONCLUSIONS": [254], "Included": [255], "showed": [257], "moderate": [258], "risk": [259], "bias,": [261], "cohorts,": [265, 331], "transparency.": [269], "excel": [274], "simulating": [276], "lack": [281], "representation.": [284], "KGs": [285, 306], "improve": [286], "top-k": [287], "support": [290], "structured": [291], "representation": [293], "reasoning.": [295], "Future": [296], "should": [298], "prioritize": [299], "hybrid": [300], "KG-chain-of-thought": [301], "architectures": [302], "integrated": [303], "open-source": [305], "UMLS": [308], "[Unified": [309], "Medical": [310], "Language": [311], "System]": [312], "SNOMED-CT": [314], "[Systematized": [315], "Nomenclature": [316], "Medicine": [318], "-": [319], "Clinical": [320], "Terms]),": [321], "parameter-efficient": [322], "fine-tuning,": [323], "compression,": [325], "multimodal": [326, 334], "LLMs,": [327], "larger": [330], "open-access": [333], "further": [337], "enhance": [338], "realism,": [339], "accuracy,": [341], "fairness,": [342], "educational": [344], "utility.": [345]}, "primary_topic": {"id": "https://openalex.org/T11636", "display_name": "Artificial Intelligence in Healthcare and Education", "score": 0.9301999807357788, "subfield": {"id": "https://openalex.org/subfields/2718", "display_name": "Health Informatics"}, "field": {"id": "https://openalex.org/fields/27", "display_name": "Medicine"}, "domain": {"id": "https://openalex.org/domains/4", "display_name": "Health Sciences"}}, "topics": [{"id": "https://openalex.org/T11636", "display_name": "Artificial Intelligence in Healthcare and Education", "score": 0.9301999807357788, "subfield": {"id": "https://openalex.org/subfields/2718", "display_name": "Health Informatics"}, "field": {"id": "https://openalex.org/fields/27", "display_name": "Medicine"}, "domain": {"id": "https://openalex.org/domains/4", "display_name": "Health Sciences"}}, {"id": "https://openalex.org/T11238", "display_name": "Simulation-Based Education in Healthcare", "score": 0.04039999842643738, "subfield": {"id": "https://openalex.org/subfields/2737", "display_name": "Physiology"}, "field": {"id": "https://openalex.org/fields/27", "display_name": "Medicine"}, "domain": {"id": "https://openalex.org/domains/4", "display_name": "Health Sciences"}}, {"id": "https://openalex.org/T10254", "display_name": "Innovations in Medical Education", "score": 0.00839999970048666, "subfield": {"id": "https://openalex.org/subfields/2739", "display_name": "Public Health, Environmental and Occupational Health"}, "field": {"id": "https://openalex.org/fields/27", "display_name": "Medicine"}, "domain": {"id": "https://openalex.org/domains/4", "display_name": "Health Sciences"}}], "primary_location": {"id": "doi:10.2196/preprints.79039", "is_oa": false, "landing_page_url": "https://doi.org/10.2196/preprints.79039", "pdf_url": null, "source": null, "license": null, "license_id": null, "version": "acceptedVersion", "is_accepted": true, "is_published": false, "raw_source_name": null, "raw_type": "posted-content"}, "cited_by_count": 0, "is_retracted": false}, {"id": "https://openalex.org/W4406785390", "doi": "https://doi.org/10.48550/arxiv.2501.13318", "title": "SplitLLM: Hierarchical Split Learning for Large Language Model over Wireless Network", "publication_year": 2025, "publication_date": "2025-01-23", "type": "preprint", "language": "en", "authorships": [{"author_position": "first", "author": {"id": "https://openalex.org/A5063892463", "display_name": "Songge Zhang", "orcid": "https://orcid.org/0000-0001-5814-1015", "observed_orcids": ["https://orcid.org/0000-0001-5814-1015"]}, "institutions": [{"id": "https://openalex.org/I20231570", "display_name": "Peking University", "ror": "https://ror.org/02v51f717", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I20231570"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Zhang, Songge", "raw_affiliation_strings": ["Peking University Shenzhen Graduate School , Peking University , Shenzhen , China The Department of Electronic and Electrical Engineering ,"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Peking University Shenzhen Graduate School , Peking University , Shenzhen , China The Department of Electronic and Electrical Engineering ,", "institution_ids": ["https://openalex.org/I20231570"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5012880878", "display_name": "Guoliang Cheng", "orcid": "https://orcid.org/0000-0001-8512-8553", "observed_orcids": ["https://orcid.org/0000-0001-8512-8553"]}, "institutions": [{"id": "https://openalex.org/I3045169105", "display_name": "Southern University of Science and Technology", "ror": "https://ror.org/049tv2d57", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I3045169105"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Cheng, Guoliang", "raw_affiliation_strings": ["Southern University of Science and Technology Department of Electronics and Information Engineering ,"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Southern University of Science and Technology Department of Electronics and Information Engineering ,", "institution_ids": ["https://openalex.org/I3045169105"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5103154174", "display_name": "Zuguang Li", "orcid": "https://orcid.org/0000-0001-8165-4775", "observed_orcids": ["https://orcid.org/0000-0001-8165-4775"]}, "institutions": [{"id": "https://openalex.org/I4210136793", "display_name": "Peng Cheng Laboratory", "ror": "https://ror.org/03qdqbt06", "country_code": "CN", "type": "facility", "lineage": ["https://openalex.org/I4210136793"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Li, Zuguang", "raw_affiliation_strings": ["Frontier Research Center , Peng Cheng Laboratory , Shenzhen , China"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Frontier Research Center , Peng Cheng Laboratory , Shenzhen , China", "institution_ids": ["https://openalex.org/I4210136793"]}]}, {"author_position": "last", "author": {"id": "https://openalex.org/A5108049214", "display_name": "Wen Wu", "orcid": "https://orcid.org/0000-0002-0458-1282", "observed_orcids": ["https://orcid.org/0000-0002-0458-1282", "https://orcid.org/0009-0007-0400-4323", "https://orcid.org/0009-0005-4678-4183"]}, "institutions": [{"id": "https://openalex.org/I204983213", "display_name": "Harbin Institute of Technology", "ror": "https://ror.org/01yqg2h08", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I204983213"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Wu, Wen", "raw_affiliation_strings": ["Harbin Institute of Technology (Shenzhen) , Shenzhen , China", "\u22c6"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Harbin Institute of Technology (Shenzhen) , Shenzhen , China", "institution_ids": ["https://openalex.org/I204983213"]}, {"raw_affiliation_string": "\u22c6", "institution_ids": []}]}], "abstract_inverted_index": {"Fine-tuning": [0], "a": [1, 76, 92], "large": [2], "language": [3], "model": [4, 104], "(LLM)": [5], "using": [6], "the": [7, 22, 42, 52, 61, 102, 116, 126, 161, 167, 181, 189, 202], "local": [8], "data": [9], "of": [10, 44, 95, 128, 154, 169], "edge": [11, 96, 148], "users": [12, 45, 48, 155], "can": [13, 192], "enable": [14], "personalized": [15], "services": [16], "and": [17, 30, 54, 68, 98, 105, 113, 119, 131, 135, 142, 150, 159, 174], "applications.": [18], "For": [19], "privacy": [20], "protection,": [21], "prevalent": [23], "solution": [24], "adopts": [25], "distributed": [26], "learning": [27, 78], "for": [28, 81, 183], "fine-tuning": [29, 82], "integrates": [31], "low-rank": [32], "adaptation": [33], "(LoRA)": [34], "to": [35, 64, 180, 198, 201], "reduce": [36, 193], "users'": [37], "computational": [38], "load.": [39], "However,": [40], "as": [41], "number": [43, 94, 153], "increases,": [46], "numerous": [47], "simultaneously": [49], "communicate": [50], "with": [51, 133], "server,": [53, 62, 91], "multiple": [55, 99], "server-side": [56], "models": [57], "concurrently": [58], "execute": [59], "on": [60], "leading": [63], "significant": [65], "communication": [66], "congestion": [67], "memory": [69, 195], "pressure.": [70], "In": [71, 144], "this": [72], "paper,": [73], "we": [74], "propose": [75], "split": [77], "(SL)": [79], "scheme": [80, 191], "LLM": [83], "in": [84, 157], "wireless": [85], "networks,": [86], "which": [87], "involves": [88], "one": [89], "cloud": [90, 182], "small": [93], "servers,": [97], "users.": [100], "Specifically,": [101], "pre-trained": [103], "LoRA": [106, 162, 176], "adapters": [107, 163, 177], "are": [108, 164, 178], "divided": [109], "into": [110], "three": [111], "parts": [112], "deployed": [114], "across": [115], "cloud,": [117, 132], "edge,": [118, 130], "user": [120], "sides.": [121], "The": [122], "training": [123], "process": [124], "follows": [125], "sequence": [127], "user,": [129], "forward": [134], "backward": [136], "propagation": [137], "achieved": [138], "by": [139], "transmitting": [140], "activation": [141], "gradient.": [143], "each": [145, 170], "round,": [146, 171], "all": [147, 172], "servers": [149], "an": [151], "equivalent": [152], "train": [156], "parallel,": [158], "only": [160], "updated.": [165], "At": [166], "end": [168], "edge-side": [173], "user-side": [175], "uploaded": [179], "aggregation.": [184], "Extensive": [185], "simulation": [186], "demonstrates": [187], "that": [188], "proposed": [190], "peak": [194], "usage": [196], "up": [197], "74%": [199], "compared": [200], "state-of-the-art": [203], "benchmarks.": [204]}, "primary_topic": {"id": "https://openalex.org/T10764", "display_name": "Privacy-Preserving Technologies in Data", "score": 0.9491999745368958, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T10764", "display_name": "Privacy-Preserving Technologies in Data", "score": 0.9491999745368958, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10273", "display_name": "IoT and Edge/Fog Computing", "score": 0.03480000048875809, "subfield": {"id": "https://openalex.org/subfields/1705", "display_name": "Computer Networks and Communications"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T11307", "display_name": "Domain Adaptation and Few-Shot Learning", "score": 0.00279999990016222, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "pmh:oai:arXiv.org:2501.13318", "is_oa": true, "landing_page_url": "http://arxiv.org/abs/2501.13318", "pdf_url": "https://arxiv.org/pdf/2501.13318", "source": {"id": "https://openalex.org/S4306400194", "display_name": "arXiv (Cornell University)", "issn_l": "2331-8422", "issn": ["2331-8422"], "is_oa": true, "is_in_doaj": false, "is_core": false, "listed_in": [], "host_organization": "https://openalex.org/I205783295", "host_organization_name": "Cornell University", "host_organization_lineage": ["https://openalex.org/I205783295"], "host_organization_lineage_names": [], "type": "repository"}, "license": "public-domain", "license_id": "https://openalex.org/licenses/public-domain", "version": "submittedVersion", "is_accepted": false, "is_published": false, "raw_source_name": null, "raw_type": "text"}, "cited_by_count": 0, "is_retracted": false}, {"id": "https://openalex.org/W4414697912", "doi": "https://doi.org/10.48550/arxiv.2509.15557", "title": "Reward Hacking Mitigation using Verifiable Composite Rewards", "publication_year": 2025, "publication_date": "2025-09-19", "type": "preprint", "language": "en", "authorships": [{"author_position": "first", "author": {"id": "https://openalex.org/A5000067159", "display_name": "Mirza Farhan Bin Tarek", "orcid": "https://orcid.org/0009-0009-0403-6442", "observed_orcids": ["https://orcid.org/0009-0009-0403-6442"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Tarek, Mirza Farhan Bin", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "last", "author": {"id": "https://openalex.org/A5044208559", "display_name": "Rahmatollah Beheshti", "orcid": "https://orcid.org/0000-0001-8912-3063", "observed_orcids": ["https://orcid.org/0000-0001-8912-3063"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Beheshti, Rahmatollah", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}], "abstract_inverted_index": {"Reinforcement": [0], "Learning": [1], "from": [2], "Verifiable": [3], "Rewards": [4], "(RLVR)": [5], "has": [6], "recently": [7], "shown": [8], "that": [9, 88], "large": [10], "language": [11], "models": [12, 125], "(LLMs)": [13], "can": [14], "develop": [15], "their": [16], "own": [17], "reasoning": [18, 40, 63, 99], "without": [19, 56], "direct": [20], "supervision.": [21], "However,": [22], "applications": [23], "in": [24], "the": [25, 39, 67, 109, 122], "medical": [26], "domain,": [27], "specifically": [28], "for": [29, 82], "question": [30], "answering,": [31], "are": [32], "susceptible": [33], "to": [34, 65, 97, 108], "significant": [35], "reward": [36, 68, 77, 94, 102, 118], "hacking": [37, 103, 119], "during": [38], "phase.": [41], "Our": [42, 85], "work": [43], "addresses": [44], "two": [45], "primary": [46], "forms": [47], "of": [48, 124], "this": [49], "behavior:": [50], "i)": [51], "providing": [52], "a": [53, 75, 114], "final": [54], "answer": [55], "preceding": [57], "reasoning,": [58], "and": [59, 104, 120], "ii)": [60], "employing": [61], "non-standard": [62], "formats": [64], "exploit": [66], "mechanism.": [69], "To": [70], "mitigate": [71], "these,": [72], "we": [73], "introduce": [74], "composite": [76], "function": [78], "with": [79, 91, 100], "specific": [80], "penalties": [81], "these": [83], "behaviors.": [84], "experiments": [86], "show": [87], "extending": [89], "RLVR": [90], "our": [92], "proposed": [93], "model": [95], "leads": [96], "better-formatted": [98], "less": [101], "good": [105], "accuracy": [106], "compared": [107], "baselines.": [110], "This": [111], "approach": [112], "marks": [113], "step": [115], "toward": [116], "reducing": [117], "enhancing": [121], "reliability": [123], "utilizing": [126], "RLVR.": [127]}, "primary_topic": {"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.4077000021934509, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.4077000021934509, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10181", "display_name": "Natural Language Processing Techniques", "score": 0.17030000686645508, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10462", "display_name": "Reinforcement Learning in Robotics", "score": 0.11869999766349792, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "pmh:oai:arXiv.org:2509.15557", "is_oa": true, "landing_page_url": "http://arxiv.org/abs/2509.15557", "pdf_url": "https://arxiv.org/pdf/2509.15557", "source": {"id": "https://openalex.org/S4306400194", "display_name": "arXiv (Cornell University)", "issn_l": "2331-8422", "issn": ["2331-8422"], "is_oa": true, "is_in_doaj": false, "is_core": false, "listed_in": [], "host_organization": "https://openalex.org/I205783295", "host_organization_name": "Cornell University", "host_organization_lineage": ["https://openalex.org/I205783295"], "host_organization_lineage_names": [], "type": "repository"}, "license": null, "license_id": null, "version": "submittedVersion", "is_accepted": false, "is_published": false, "raw_source_name": null, "raw_type": "text"}, "cited_by_count": 0, "is_retracted": false}, {"id": "https://openalex.org/W4417298726", "doi": "https://doi.org/10.48550/arxiv.2505.13844", "title": "Improve Language Model and Brain Alignment via Associative Memory", "publication_year": 2025, "publication_date": "2025-05-20", "type": "preprint", "language": "en", "authorships": [{"author_position": "first", "author": {"id": "https://openalex.org/A5057105022", "display_name": "Congchi Yin", "orcid": "https://orcid.org/0000-0002-7381-0851", "observed_orcids": ["https://orcid.org/0000-0002-7381-0851"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Yin, Congchi", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5002568596", "display_name": "Yongpeng Zhang", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Zhang, Yongpeng", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5073301735", "display_name": "Xuyun Wen", "orcid": "https://orcid.org/0000-0003-2230-8658", "observed_orcids": ["https://orcid.org/0000-0003-2230-8658"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Wen, Xuyun", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "last", "author": {"id": "https://openalex.org/A5061435467", "display_name": "Piji Li", "orcid": "https://orcid.org/0000-0003-1474-3692", "observed_orcids": ["https://orcid.org/0000-0003-1474-3692"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Li, Piji", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}], "abstract_inverted_index": {"Associative": [0], "memory": [1, 63, 90, 121], "engages": [2], "in": [3, 11, 83], "the": [4, 12, 40, 55, 74, 109], "integration": [5], "of": [6, 115], "relevant": [7], "information": [8, 33], "for": [9], "comprehension": [10], "human": [13, 28], "cognition": [14], "system.": [15], "In": [16], "this": [17], "work,": [18], "we": [19], "seek": [20], "to": [21, 52, 68, 88], "improve": [22], "alignment": [23, 41, 75], "between": [24, 42, 76], "language": [25, 43, 49, 70, 77, 96], "models": [26, 97], "and": [27, 45, 79, 124], "brain": [29, 46, 53, 80, 84, 105], "while": [30], "processing": [31], "speech": [32], "by": [34, 47, 107], "integrating": [35], "associative": [36, 62, 89, 120], "memory.": [37], "After": [38], "verifying": [39], "model": [44, 50, 78], "mapping": [48], "activations": [51], "activity,": [54], "original": [56], "text": [57], "stimuli": [58], "expanded": [59], "with": [60, 104, 117], "simulated": [61], "are": [64], "regarded": [65], "as": [66, 122, 127], "input": [67, 123], "computational": [69], "models.": [71], "We": [72, 92], "find": [73], "is": [81], "improved": [82], "regions": [85], "closely": [86], "related": [87], "processing.": [91], "also": [93], "demonstrate": [94], "large": [95], "after": [98], "specific": [99], "supervised": [100], "fine-tuning": [101], "better": [102], "align": [103], "response,": [106], "building": [108], "\\textit{Association}": [110], "dataset": [111], "containing": [112], "1000": [113], "samples": [114], "stories,": [116], "instructions": [118], "encouraging": [119], "associated": [125], "content": [126], "output.": [128]}, "primary_topic": {"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.48969998955726624, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.48969998955726624, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10465", "display_name": "Neurobiology of Language and Bilingualism", "score": 0.21250000596046448, "subfield": {"id": "https://openalex.org/subfields/2805", "display_name": "Cognitive Neuroscience"}, "field": {"id": "https://openalex.org/fields/28", "display_name": "Neuroscience"}, "domain": {"id": "https://openalex.org/domains/1", "display_name": "Life Sciences"}}, {"id": "https://openalex.org/T11714", "display_name": "Multimodal Machine Learning Applications", "score": 0.0697999969124794, "subfield": {"id": "https://openalex.org/subfields/1707", "display_name": "Computer Vision and Pattern Recognition"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "pmh:oai:arXiv.org:2505.13844", "is_oa": true, "landing_page_url": "http://arxiv.org/abs/2505.13844", "pdf_url": "https://arxiv.org/pdf/2505.13844", "source": {"id": "https://openalex.org/S4306400194", "display_name": "arXiv (Cornell University)", "issn_l": "2331-8422", "issn": ["2331-8422"], "is_oa": true, "is_in_doaj": false, "is_core": false, "listed_in": [], "host_organization": "https://openalex.org/I205783295", "host_organization_name": "Cornell University", "host_organization_lineage": ["https://openalex.org/I205783295"], "host_organization_lineage_names": [], "type": "repository"}, "license": null, "license_id": null, "version": "submittedVersion", "is_accepted": false, "is_published": false, "raw_source_name": null, "raw_type": "text"}, "cited_by_count": 0, "is_retracted": false}, {"id": "https://openalex.org/W7106845689", "doi": "https://doi.org/10.48550/arxiv.2511.19654", "title": "Accuracy and Efficiency Trade-Offs in LLM-Based Malware Detection and Explanation: A Comparative Study of Parameter Tuning vs. Full Fine-Tuning", "publication_year": 2025, "publication_date": "2025-11-24", "type": "preprint", "language": null, "authorships": [{"author_position": "first", "author": {"id": null, "display_name": "Gravereaux, Stephen C.", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Gravereaux, Stephen C.", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "last", "author": {"id": null, "display_name": "Islam, Sheikh Rabiul", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Islam, Sheikh Rabiul", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}], "abstract_inverted_index": {"This": [0], "study": [1], "examines": [2], "whether": [3], "Low-Rank": [4], "Adaptation": [5], "(LoRA)": [6], "fine-tuned": [7, 18, 73], "Large": [8], "Language": [9], "Models": [10], "(LLMs)": [11], "can": [12], "approximate": [13], "the": [14, 81], "performance": [15, 103], "of": [16, 90, 140], "fully": [17, 72], "models": [19, 100], "in": [20, 147, 172], "generating": [21], "human-interpretable": [22], "decisions": [23], "and": [24, 58, 70, 87, 117, 142, 169], "explanations": [25, 159], "for": [26, 54, 160], "malware": [27, 31, 161, 173], "classification.": [28], "Achieving": [29], "trustworthy": [30], "detection,": [32], "particularly": [33], "when": [34], "LLMs": [35], "are": [36], "involved,": [37], "remains": [38], "a": [39, 71, 124, 137], "significant": [40], "challenge.": [41], "We": [42], "developed": [43], "an": [44], "evaluation": [45], "framework": [46], "using": [47], "Bilingual": [48], "Evaluation": [49, 56], "Understudy": [50, 53], "(BLEU),": [51], "Recall-Oriented": [52], "Gisting": [55], "(ROUGE),": [57], "Semantic": [59], "Similarity": [60], "Metrics": [61], "to": [62, 92], "benchmark": [63], "explanation": [64, 152], "quality": [65], "across": [66], "five": [67], "LoRA": [68, 95, 99, 125, 135], "configurations": [69], "baseline.": [74], "Results": [75], "indicate": [76], "that": [77, 134], "full": [78, 105], "fine-tuning": [79, 106], "achieves": [80], "highest": [82], "overall": [83], "scores,": [84], "with": [85, 127], "BLEU": [86], "ROUGE": [88], "improvements": [89], "up": [91], "10%": [93], "over": [94, 121], "variants.": [96], "However,": [97], "mid-range": [98], "deliver": [101], "competitive": [102], "exceeding": [104], "on": [107, 123], "two": [108], "metrics": [109], "while": [110], "reducing": [111], "model": [112, 126], "size": [113], "by": [114, 120], "approximately": [115], "81%": [116], "training": [118], "time": [119], "80%": [122], "15.5%": [128], "trainable": [129], "parameters.": [130], "These": [131], "findings": [132], "demonstrate": [133], "offers": [136], "practical": [138], "balance": [139], "interpretability": [141], "resource": [143], "efficiency,": [144], "enabling": [145], "deployment": [146], "resource-constrained": [148], "environments": [149], "without": [150], "sacrificing": [151], "quality.": [153], "By": [154], "providing": [155], "feature-driven": [156], "natural": [157], "language": [158], "classifications,": [162], "this": [163], "approach": [164], "enhances": [165], "transparency,": [166], "analyst": [167], "confidence,": [168], "operational": [170], "scalability": [171], "detection": [174], "systems.": [175]}, "primary_topic": {"id": "https://openalex.org/T11241", "display_name": "Advanced Malware Detection Techniques", "score": 0.9717000126838684, "subfield": {"id": "https://openalex.org/subfields/1711", "display_name": "Signal Processing"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T11241", "display_name": "Advanced Malware Detection Techniques", "score": 0.9717000126838684, "subfield": {"id": "https://openalex.org/subfields/1711", "display_name": "Signal Processing"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T12026", "display_name": "Explainable Artificial Intelligence (XAI)", "score": 0.020400000736117363, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T11689", "display_name": "Adversarial Robustness in Machine Learning", "score": 0.0031999999191612005, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "doi:10.48550/arxiv.2511.19654", "is_oa": true, "landing_page_url": "https://doi.org/10.48550/arxiv.2511.19654", "pdf_url": null, "source": {"id": "https://openalex.org/S4306400194", "display_name": "arXiv (Cornell University)", "issn_l": "2331-8422", "issn": ["2331-8422"], "is_oa": true, "is_in_doaj": false, "is_core": false, "listed_in": [], "host_organization": "https://openalex.org/I205783295", "host_organization_name": "Cornell University", "host_organization_lineage": ["https://openalex.org/I205783295"], "host_organization_lineage_names": [], "type": "repository"}, "license": "cc-by", "license_id": "https://openalex.org/licenses/cc-by", "version": null, "is_accepted": false, "is_published": false, "raw_source_name": null, "raw_type": "Preprint"}, "cited_by_count": 0, "is_retracted": false}, {"id": "https://openalex.org/W4406093146", "doi": "https://doi.org/10.22028/d291-47482", "title": "SaLoRA: Safety-Alignment Preserved Low-Rank Adaptation", "publication_year": 2025, "publication_date": "2025-01-01", "type": "preprint", "language": "en", "authorships": [{"author_position": "first", "author": {"id": "https://openalex.org/A5100416259", "display_name": "Mingjie Li", "orcid": "https://orcid.org/0000-0002-1588-2654", "observed_orcids": ["https://orcid.org/0000-0002-1588-2654", "https://orcid.org/0000-0002-2989-2669"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Li, Mingjie", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5013107086", "display_name": "Wai Man Si", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Si, Wai Man", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5102730269", "display_name": "Michael Backes", "orcid": "https://orcid.org/0000-0002-7130-9211", "observed_orcids": ["https://orcid.org/0000-0002-7130-9211"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Backes, Michael", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5040809339", "display_name": "Y Victoria Zhang", "orcid": "https://orcid.org/0000-0003-4496-2650", "observed_orcids": ["https://orcid.org/0000-0003-4496-2650", "https://orcid.org/0000-0002-1083-459X", "https://orcid.org/0000-0002-2522-2027", "https://orcid.org/0000-0003-1217-3876", "https://orcid.org/0009-0001-5370-4381", "https://orcid.org/0009-0004-2024-2434", "https://orcid.org/0009-0004-4860-1910", "https://orcid.org/0009-0007-1388-3737", "https://orcid.org/0009-0006-3503-7593"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Zhang, Yang", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "last", "author": {"id": "https://openalex.org/A5004999983", "display_name": "Yisen Wang", "orcid": "https://orcid.org/0000-0002-1612-4433", "observed_orcids": ["https://orcid.org/0000-0002-1612-4433"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Wang, Yisen", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}], "abstract_inverted_index": {"As": [0], "advancements": [1], "in": [2, 28, 48, 70, 97, 137], "large": [3], "language": [4], "models": [5, 13], "(LLMs)": [6], "continue": [7], "and": [8, 76, 90, 110], "the": [9, 45, 54, 63, 68], "demand": [10], "for": [11, 53, 94], "personalized": [12], "increases,": [14], "parameter-efficient": [15], "fine-tuning": [16, 41, 139], "(PEFT)": [17], "methods": [18, 109], "(e.g.,": [19], "LoRA)": [20], "will": [21], "become": [22], "essential": [23], "due": [24], "to": [25, 117], "their": [26, 111, 121], "efficiency": [27], "reducing": [29], "computation": [30], "costs.": [31], "However,": [32], "recent": [33], "studies": [34], "have": [35], "raised": [36], "alarming": [37], "concerns": [38], "that": [39, 127], "LoRA": [40, 108], "could": [42], "potentially": [43], "compromise": [44], "safety": [46, 71, 84, 88], "alignment": [47, 72], "LLMs,": [49], "posing": [50], "significant": [51], "risks": [52], "model": [55], "owner.": [56], "In": [57], "this": [58], "paper,": [59], "we": [60, 80], "first": [61], "investigate": [62], "underlying": [64], "mechanism": [65], "by": [66, 87], "analyzing": [67], "changes": [69], "related": [73], "features": [74], "before": [75], "after": [77], "fine-tuning.": [78], "Then,": [79], "propose": [81], "a": [82, 91], "fixed": [83], "module": [85], "calculated": [86], "data": [89], "task-specific": [92], "initialization": [93], "trainable": [95], "parameters": [96], "low-rank": [98], "adaptations,": [99], "termed": [100], "Safety-alignment": [101], "preserved": [102], "Low-Rank": [103], "Adaptation": [104], "(SaLoRA).": [105], "Unlike": [106], "previous": [107], "variants,": [112], "SaLoRA": [113, 128], "enables": [114], "targeted": [115], "modifications": [116], "LLMs": [118], "without": [119], "disrupting": [120], "original": [122], "alignments.": [123], "Our": [124], "experiments": [125], "show": [126], "outperforms": [129], "various": [130, 134], "adapters-based": [131], "approaches": [132], "across": [133], "evaluation": [135], "metrics": [136], "different": [138], "tasks.": [140]}, "primary_topic": {"id": "https://openalex.org/T11307", "display_name": "Domain Adaptation and Few-Shot Learning", "score": 0.6338000297546387, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T11307", "display_name": "Domain Adaptation and Few-Shot Learning", "score": 0.6338000297546387, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.17900000512599945, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T11689", "display_name": "Adversarial Robustness in Machine Learning", "score": 0.10019999742507935, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "pmh:oai:arXiv.org:2501.01765", "is_oa": true, "landing_page_url": "http://arxiv.org/abs/2501.01765", "pdf_url": "https://arxiv.org/pdf/2501.01765", "source": {"id": "https://openalex.org/S4306400194", "display_name": "arXiv (Cornell University)", "issn_l": "2331-8422", "issn": ["2331-8422"], "is_oa": true, "is_in_doaj": false, "is_core": false, "listed_in": [], "host_organization": "https://openalex.org/I205783295", "host_organization_name": "Cornell University", "host_organization_lineage": ["https://openalex.org/I205783295"], "host_organization_lineage_names": [], "type": "repository"}, "license": "cc-by", "license_id": "https://openalex.org/licenses/cc-by", "version": "submittedVersion", "is_accepted": false, "is_published": false, "raw_source_name": null, "raw_type": "text"}, "cited_by_count": 1, "is_retracted": false}, {"id": "https://openalex.org/W4416342072", "doi": "https://doi.org/10.3389/frai.2025.1681277", "title": "Resource-efficient fine-tuning of large vision-language models for multimodal perception in autonomous excavators", "publication_year": 2025, "publication_date": "2025-11-18", "type": "article", "language": "en", "authorships": [{"author_position": "first", "author": {"id": "https://openalex.org/A5084971994", "display_name": "Hung Viet Nguyen", "orcid": "https://orcid.org/0000-0002-6981-0891", "observed_orcids": ["https://orcid.org/0000-0002-6981-0891"]}, "institutions": [{"id": "https://openalex.org/I104338594", "display_name": "Inje University", "ror": "https://ror.org/04xqwq985", "country_code": "KR", "type": "education", "lineage": ["https://openalex.org/I104338594"]}], "countries": ["KR"], "is_corresponding": false, "raw_author_name": "Hung Viet Nguyen", "raw_affiliation_strings": ["Department of Digital Anti-aging Healthcare, INJE University, Kimhae, Republic of Korea"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Department of Digital Anti-aging Healthcare, INJE University, Kimhae, Republic of Korea", "institution_ids": ["https://openalex.org/I104338594"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5020627604", "display_name": "Hyo-Jin Park", "orcid": "https://orcid.org/0000-0001-5021-4240", "observed_orcids": ["https://orcid.org/0000-0001-5021-4240"]}, "institutions": [{"id": "https://openalex.org/I71007489", "display_name": "Kyungnam University", "ror": "https://ror.org/037pkxm09", "country_code": "KR", "type": "education", "lineage": ["https://openalex.org/I71007489"]}], "countries": ["KR"], "is_corresponding": false, "raw_author_name": "Hyojin Park", "raw_affiliation_strings": ["Gyeongnam Intelligence Innovation Center (GIIC), Kyungnam University, Changwon, Republic of Korea"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Gyeongnam Intelligence Innovation Center (GIIC), Kyungnam University, Changwon, Republic of Korea", "institution_ids": ["https://openalex.org/I71007489"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5016249429", "display_name": "Nam\u2010Hyun Yoo", "orcid": "https://orcid.org/0000-0003-2161-9251", "observed_orcids": ["https://orcid.org/0000-0003-2161-9251"]}, "institutions": [{"id": "https://openalex.org/I71007489", "display_name": "Kyungnam University", "ror": "https://ror.org/037pkxm09", "country_code": "KR", "type": "education", "lineage": ["https://openalex.org/I71007489"]}], "countries": ["KR"], "is_corresponding": false, "raw_author_name": "Namhyun Yoo", "raw_affiliation_strings": ["Department of Computer Engineering, Kyungnam University, Changwon, Republic of Korea"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Department of Computer Engineering, Kyungnam University, Changwon, Republic of Korea", "institution_ids": ["https://openalex.org/I71007489"]}]}, {"author_position": "last", "author": {"id": "https://openalex.org/A5114860745", "display_name": "Jinhong Yang", "orcid": "https://orcid.org/0000-0002-7756-0263", "observed_orcids": ["https://orcid.org/0000-0002-7756-0263"]}, "institutions": [{"id": "https://openalex.org/I104338594", "display_name": "Inje University", "ror": "https://ror.org/04xqwq985", "country_code": "KR", "type": "education", "lineage": ["https://openalex.org/I104338594"]}], "countries": ["KR"], "is_corresponding": true, "raw_author_name": "Jinhong Yang", "raw_affiliation_strings": ["Department of Digital Anti-aging Healthcare, INJE University, Kimhae, Republic of Korea", "Department of Medical IT, INJE University, Kimhae, Republic of Korea"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Department of Digital Anti-aging Healthcare, INJE University, Kimhae, Republic of Korea", "institution_ids": ["https://openalex.org/I104338594"]}, {"raw_affiliation_string": "Department of Medical IT, INJE University, Kimhae, Republic of Korea", "institution_ids": ["https://openalex.org/I104338594"]}]}], "abstract_inverted_index": {"Recent": [0], "advances": [1], "in": [2, 42, 90, 147, 209], "large": [3], "vision-language": [4], "models": [5], "(LVLMs)": [6], "have": [7], "transformed": [8], "visual": [9, 31, 40], "recognition": [10, 41], "research": [11], "by": [12], "enabling": [13], "multimodal": [14, 196], "integration": [15], "of": [16, 30, 36, 70, 78, 159, 162, 165, 170, 193], "images,": [17], "text,": [18], "and": [19, 26, 72, 102, 126, 137, 151, 167, 180, 206], "videos.": [20], "This": [21], "fusion": [22], "supports": [23], "a": [24, 53, 65, 112], "deeper": [25], "more": [27], "context-aware": [28], "understanding": [29], "environments.": [32], "However,": [33], "the": [34, 93, 191], "application": [35], "LVLMs": [37, 58], "to": [38, 60], "multitask": [39], "real-world": [43], "construction": [44], "scenarios": [45], "remains": [46], "underexplored.": [47], "In": [48], "this": [49], "study,": [50], "we": [51], "present": [52], "resource-efficient": [54], "framework": [55], "for": [56, 199], "fine-tuning": [57, 104], "tailored": [59], "autonomous": [61, 210], "excavator": [62, 211], "operations,": [63], "with": [64, 92, 106, 154], "focus": [66], "on": [67, 81, 133, 139], "robust": [68], "detection": [69, 150], "humans": [71, 179], "obstacles,": [73], "as": [74, 76], "well": [75], "classification": [77], "weather": [79, 152, 186], "conditions": [80], "consumer-grade": [82], "hardware.": [83], "By": [84], "leveraging": [85], "Quantized": [86], "Low-Rank": [87], "Adaptation": [88], "(QLoRA)": [89], "conjunction": [91], "Unsloth": [94], "framework,": [95], "our": [96], "method": [97], "substantially": [98], "reduces": [99], "memory": [100], "consumption": [101], "accelerates": [103], "compared": [105], "conventional": [107], "approaches.": [108], "We": [109], "comprehensively": [110], "evaluate": [111], "domain-specific": [113], "excavator-vision": [114], "dataset": [115], "using": [116], "five": [117], "open-source": [118], "LVLMs.": [119], "These": [120, 188], "include": [121], "Llama-3.2-Vision,": [122], "Qwen2-VL,": [123], "Qwen2.5-VL,": [124], "LLaVA-1.6,": [125], "Gemma": [127], "3.": [128], "Each": [129], "model": [130, 175], "is": [131], "fine-tuned": [132, 173], "1,000": [134], "annotated": [135], "frames": [136], "tested": [138], "2000": [140], "images.": [141], "Experimental": [142], "results": [143, 189], "demonstrate": [144], "significant": [145], "improvements": [146], "both": [148], "object": [149], "classification,": [153], "Qwen2-VL-7B": [155, 174], "achieving": [156], "an": [157], "mAP@50": [158], "88.03%,": [160], "mAP@[0.50:0.95]": [161], "74.20%,": [163], "accuracy": [164], "84.54%,": [166], "F1": [168], "score": [169], "78.83%.": [171], "Our": [172], "not": [176], "only": [177], "detects": [178], "obstacles": [181], "robustly": [182], "but": [183], "also": [184], "classifies": [185], "accurately.": [187], "illustrate": [190], "feasibility": [192], "deploying": [194], "LVLM-based": [195], "AI": [197], "agents": [198], "safety": [200], "monitoring,": [201], "pose": [202], "estimation,": [203], "activity": [204], "tracking,": [205], "strategic": [207], "planning": [208], "operations.": [212]}, "primary_topic": {"id": "https://openalex.org/T11714", "display_name": "Multimodal Machine Learning Applications", "score": 0.6187000274658203, "subfield": {"id": "https://openalex.org/subfields/1707", "display_name": "Computer Vision and Pattern Recognition"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T11714", "display_name": "Multimodal Machine Learning Applications", "score": 0.6187000274658203, "subfield": {"id": "https://openalex.org/subfields/1707", "display_name": "Computer Vision and Pattern Recognition"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10036", "display_name": "Advanced Neural Network Applications", "score": 0.35740000009536743, "subfield": {"id": "https://openalex.org/subfields/1707", "display_name": "Computer Vision and Pattern Recognition"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T11307", "display_name": "Domain Adaptation and Few-Shot Learning", "score": 0.011900000274181366, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "doi:10.3389/frai.2025.1681277", "is_oa": true, "landing_page_url": "https://doi.org/10.3389/frai.2025.1681277", "pdf_url": "https://public-pages-files-2025.frontiersin.org/journals/artificial-intelligence/articles/10.3389/frai.2025.1681277/pdf", "source": {"id": "https://openalex.org/S4210197006", "display_name": "Frontiers in Artificial Intelligence", "issn_l": "2624-8212", "issn": ["2624-8212"], "is_oa": true, "is_in_doaj": true, "is_core": true, "listed_in": ["cwts-core", "doaj", "jufo-1", "ki-jl-1", "norway-1"], "host_organization": "https://openalex.org/P4310320527", "host_organization_name": "Frontiers Media", "host_organization_lineage": ["https://openalex.org/P4310320527"], "host_organization_lineage_names": ["Frontiers Media"], "type": "journal"}, "license": "cc-by", "license_id": "https://openalex.org/licenses/cc-by", "version": "publishedVersion", "is_accepted": true, "is_published": true, "raw_source_name": "Frontiers in Artificial Intelligence", "raw_type": "journal-article"}, "cited_by_count": 2, "is_retracted": false}, {"id": "https://openalex.org/W7106476231", "doi": null, "title": "DEPO: Dual-Efficiency Preference Optimization for LLM Agents", "publication_year": 2025, "publication_date": "2025-11-19", "type": "preprint", "language": null, "authorships": [{"author_position": "first", "author": {"id": null, "display_name": "Chen, Sirui", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Chen, Sirui", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": null, "display_name": "Zhao, Mengshi", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Zhao, Mengshi", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": null, "display_name": "Xu, Lei", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Xu, Lei", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": null, "display_name": "Zhao, Yuying", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Zhao, Yuying", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": null, "display_name": "Zhu, Beier", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Zhu, Beier", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": null, "display_name": "Zhang, Hanwang", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Zhang, Hanwang", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": null, "display_name": "Zhao, Shengjie", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Zhao, Shengjie", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "last", "author": {"id": null, "display_name": "Lu, Chaochao", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Lu, Chaochao", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}], "abstract_inverted_index": {"Recent": [0], "advances": [1], "in": [2, 36, 130], "large": [3], "language": [4], "models": [5], "(LLMs)": [6], "have": [7], "greatly": [8], "improved": [9], "their": [10], "reasoning": [11], "and": [12, 67, 98, 105, 117, 140], "decision-making": [13], "abilities": [14], "when": [15, 145], "deployed": [16], "as": [17], "agents.": [18], "Richer": [19], "reasoning,": [20], "however,": [21], "often": [22], "comes": [23], "at": [24, 157], "the": [25, 73, 151], "cost": [26], "of": [27, 30, 45, 75, 150], "longer": [28], "chain": [29], "thought": [31], "(CoT),": [32], "hampering": [33], "interaction": [34], "efficiency": [35, 143], "real-world": [37], "scenarios.": [38], "Nevertheless,": [39], "there": [40], "still": [41], "lacks": [42], "systematic": [43], "definition": [44], "LLM": [46], "agent": [47], "efficiency,": [48, 61, 70], "hindering": [49], "targeted": [50], "improvements.": [51], "To": [52], "this": [53, 83], "end,": [54], "we": [55, 85], "introduce": [56], "dual-efficiency,": [57], "comprising": [58], "(i)": [59], "step-level": [60], "which": [62, 71], "minimizes": [63, 72], "tokens": [64], "per": [65], "step,": [66], "(ii)": [68], "trajectory-level": [69], "number": [74], "steps": [76, 118], "to": [77, 115, 121, 126, 135], "complete": [78], "a": [79, 88, 127], "task.": [80], "Building": [81], "on": [82, 103, 147], "definition,": [84], "propose": [86], "DEPO,": [87], "dual-efficiency": [89], "preference": [90], "optimization": [91], "method": [92], "that": [93, 108], "jointly": [94], "rewards": [95], "succinct": [96], "responses": [97], "fewer": [99], "action": [100], "steps.": [101], "Experiments": [102], "WebShop": [104], "BabyAI": [106], "show": [107], "DEPO": [109, 132], "cuts": [110], "token": [111], "usage": [112], "by": [113, 119], "up": [114, 120, 125], "60.9%": [116], "26.9%,": [122], "while": [123], "achieving": [124], "29.3%": [128], "improvement": [129], "performance.": [131], "also": [133], "generalizes": [134], "three": [136], "out-of-domain": [137], "math": [138], "benchmarks": [139], "retains": [141], "its": [142], "gains": [144], "trained": [146], "only": [148], "25%": [149], "data.": [152], "Our": [153], "project": [154], "page": [155], "is": [156], "https://opencausalab.github.io/DEPO.": [158]}, "primary_topic": {"id": "https://openalex.org/T10906", "display_name": "AI-based Problem Solving and Planning", "score": 0.4066999852657318, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T10906", "display_name": "AI-based Problem Solving and Planning", "score": 0.4066999852657318, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10462", "display_name": "Reinforcement Learning in Robotics", "score": 0.1525000035762787, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10181", "display_name": "Natural Language Processing Techniques", "score": 0.13230000436306, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "pmh:oai:arXiv.org:2511.15392", "is_oa": true, "landing_page_url": "http://arxiv.org/abs/2511.15392", "pdf_url": "https://arxiv.org/pdf/2511.15392", "source": {"id": "https://openalex.org/S4306400194", "display_name": "arXiv (Cornell University)", "issn_l": "2331-8422", "issn": ["2331-8422"], "is_oa": true, "is_in_doaj": false, "is_core": false, "listed_in": [], "host_organization": "https://openalex.org/I205783295", "host_organization_name": "Cornell University", "host_organization_lineage": ["https://openalex.org/I205783295"], "host_organization_lineage_names": [], "type": "repository"}, "license": null, "license_id": null, "version": "submittedVersion", "is_accepted": false, "is_published": false, "raw_source_name": null, "raw_type": "text"}, "cited_by_count": 0, "is_retracted": false}, {"id": "https://openalex.org/W4407806684", "doi": "https://doi.org/10.48550/arxiv.2502.14289", "title": "Drift: Decoding-time Personalized Alignments with Implicit User Preferences", "publication_year": 2025, "publication_date": "2025-02-20", "type": "preprint", "language": "en", "authorships": [{"author_position": "first", "author": {"id": "https://openalex.org/A5012270962", "display_name": "Minbeom Kim", "orcid": "https://orcid.org/0000-0002-3879-1093", "observed_orcids": ["https://orcid.org/0000-0002-3879-1093"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Kim, Minbeom", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5023464943", "display_name": "Kang-il Lee", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Lee, Kang-il", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": null, "display_name": "Joo, Seongho", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Joo, Seongho", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5054334046", "display_name": "Hwaran Lee", "orcid": "https://orcid.org/0000-0002-3773-4871", "observed_orcids": ["https://orcid.org/0000-0002-3773-4871"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Lee, Hwaran", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": null, "display_name": "Thonet, Thibaut", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Thonet, Thibaut", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "last", "author": {"id": null, "display_name": "Jung, Kyomin", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Jung, Kyomin", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}], "abstract_inverted_index": {"Personalized": [0], "alignments": [1], "for": [2], "individual": [3], "users": [4], "have": [5], "been": [6], "a": [7, 18, 53, 58, 64, 77, 96, 102], "long-standing": [8], "goal": [9], "in": [10, 52], "large": [11], "language": [12], "models": [13, 73], "(LLMs).": [14], "We": [15], "introduce": [16], "Drift,": [17], "novel": [19], "framework": [20], "that": [21, 108, 124], "personalizes": [22, 50], "LLMs": [23, 51], "at": [24, 86], "decoding": [25, 87], "time": [26, 88], "with": [27], "implicit": [28], "user": [29, 74], "preferences.": [30], "Traditional": [31], "Reinforcement": [32], "Learning": [33], "from": [34], "Human": [35], "Feedback": [36], "(RLHF)": [37], "requires": [38], "thousands": [39], "of": [40, 79], "annotated": [41], "examples": [42, 61], "and": [43, 83, 101, 121, 130], "expensive": [44], "gradient": [45], "updates.": [46], "In": [47], "contrast,": [48], "Drift": [49, 109, 125], "training-free": [54], "manner,": [55], "using": [56, 115], "only": [57, 116], "few": [59], "dozen": [60], "to": [62, 89], "steer": [63], "frozen": [65], "model": [66], "through": [67], "efficient": [68, 129], "preference": [69], "modeling.": [70], "Our": [71, 119], "approach": [72], "preferences": [75], "as": [76], "composition": [78], "predefined,": [80], "interpretable": [81], "attributes": [82], "aligns": [84], "them": [85], "enable": [90], "personalized": [91], "generation.": [92], "Experiments": [93], "on": [94], "both": [95, 127], "synthetic": [97], "persona": [98], "dataset": [99, 105], "(Perspective)": [100], "real": [103], "human-annotated": [104], "(PRISM)": [106], "demonstrate": [107], "significantly": [110], "outperforms": [111], "RLHF": [112], "baselines": [113], "while": [114], "50-100": [117], "examples.": [118], "results": [120], "analysis": [122], "show": [123], "is": [126], "computationally": [128], "interpretable.": [131]}, "primary_topic": {"id": "https://openalex.org/T10181", "display_name": "Natural Language Processing Techniques", "score": 0.3463999927043915, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T10181", "display_name": "Natural Language Processing Techniques", "score": 0.3463999927043915, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.20999999344348907, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T12031", "display_name": "Speech and dialogue systems", "score": 0.18700000643730164, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "pmh:oai:arXiv.org:2502.14289", "is_oa": true, "landing_page_url": "http://arxiv.org/abs/2502.14289", "pdf_url": "https://arxiv.org/pdf/2502.14289", "source": {"id": "https://openalex.org/S4306400194", "display_name": "arXiv (Cornell University)", "issn_l": "2331-8422", "issn": ["2331-8422"], "is_oa": true, "is_in_doaj": false, "is_core": false, "listed_in": [], "host_organization": "https://openalex.org/I205783295", "host_organization_name": "Cornell University", "host_organization_lineage": ["https://openalex.org/I205783295"], "host_organization_lineage_names": [], "type": "repository"}, "license": null, "license_id": null, "version": "submittedVersion", "is_accepted": false, "is_published": false, "raw_source_name": null, "raw_type": "text"}, "cited_by_count": 0, "is_retracted": false}, {"id": "https://openalex.org/W4416445771", "doi": "https://doi.org/10.36227/techrxiv.176369747.70179230/v1", "title": "Efficient Long Chain-of-Thought Elicitation through Synthetic Data Generation and Targeted Fine-Tuning", "publication_year": 2025, "publication_date": "2025-11-21", "type": "preprint", "language": null, "authorships": [{"author_position": "first", "author": {"id": "https://openalex.org/A5042342823", "display_name": "Abdulloh Abdulloh", "orcid": "https://orcid.org/0009-0003-7300-5638", "observed_orcids": ["https://orcid.org/0009-0003-7300-5638"]}, "institutions": [], "countries": [], "is_corresponding": true, "raw_author_name": "Abdulloh Abdulloh", "raw_affiliation_strings": ["Independent Researcher"], "raw_orcid": "https://orcid.org/0009-0003-7300-5638", "affiliations": [{"raw_affiliation_string": "Independent Researcher", "institution_ids": []}]}], "abstract_inverted_index": {"Recent": [0], "advances": [1], "in": [2, 124, 270, 314], "large": [3, 107], "language": [4, 126], "model": [5, 249], "(LLM)": [6], "reasoning": [7, 18, 122, 155, 207, 213, 278], "have": [8], "demonstrated": [9], "that": [10, 74, 147, 189, 223, 292], "scaling": [11], "inference-time": [12], "compute": [13], "through": [14, 103], "long": [15, 31, 80, 87, 120, 149, 182, 284, 316], "chain-of-thought": [16], "(CoT)": [17], "can": [19], "significantly": [20], "improve": [21], "performance": [22, 230, 286], "on": [23, 36, 79, 211, 252, 257], "complex": [24], "tasks.": [25], "However,": [26], "current": [27], "approaches": [28], "for": [29, 118, 262, 283, 299], "training": [30, 89, 133], "CoT": [32, 81, 88, 121, 150, 164, 183, 285, 317], "capabilities": [33, 123], "rely": [34], "heavily": [35], "either": [37], "(1)": [38, 169], "expensive": [39], "reinforcement": [40], "learning": [41, 203, 295], "(RL)": [42], "from": [43, 53, 106, 158, 172], "base": [44], "models": [45, 56, 127, 306], "requiring": [46, 235], "massive": [47], "computational": [48, 238], "resources,": [49], "or": [50, 134], "(2)": [51, 185], "distillation": [52], "proprietary": [54, 244], "frontier": [55, 137], "like": [57], "OpenAI\u2019s": [58], "o1,": [59], "which": [60, 277], "is": [61, 297], "neither": [62], "reproducible": [63, 312], "nor": [64], "accessible": [65, 315], "to": [66, 94, 136, 178, 204, 231, 243, 260, 310], "most": [67, 281], "researchers.": [68], "Furthermore,": [69], "existing": [70, 159], "work": [71], "has": [72], "shown": [73], "while": [75, 234], "supervised": [76], "fine-tuning": [77], "(SFT)": [78], "data": [82, 90, 144, 161, 226, 254], "improves": [83], "performance,": [84], "obtaining": [85], "high-quality": [86], "remains": [91], "challenging": [92], "due": [93], "limited": [95], "availability": [96], "and": [97, 162, 175, 194, 197, 219, 240, 265, 287, 305], "the": [98], "expense": [99], "of": [100, 272], "generating": [101], "it": [102], "rejection": [104], "sampling": [105], "teacher": [108], "models.": [109, 138, 245], "In": [110], "this": [111], "work,": [112], "we": [113, 221], "investigate": [114], "computationally": [115], "efficient": [116], "methods": [117, 233], "eliciting": [119], "smaller": [125], "(\u22648B": [128], "parameters)": [129], "without": [130], "extensive": [131], "RL": [132], "access": [135, 242], "We": [139, 274], "propose": [140], "a": [141], "novel": [142], "synthetic": [143, 225, 253], "generation": [145], "framework": [146], "constructs": [148], "trajectories": [151], "by": [152], "programmatically": [153], "composing": [154], "patterns": [156, 279], "extracted": [157], "web": [160], "short": [163], "examples.": [165], "Our": [166, 246], "approach": [167, 227, 296], "combines": [168], "pattern": [170], "mining": [171], "discussion": [173], "forums": [174], "educational": [176], "content": [177], "identify": [179], "naturally": [180], "occurring": [181], "structures,": [184], "rule-based": [186], "trajectory": [187], "construction": [188], "systematically": [190], "incorporates": [191], "self-correction,": [192], "branching,": [193], "verification": [195], "steps,": [196], "(3)": [198], "targeted": [199], "SFT": [200], "with": [201], "curriculum": [202, 294], "progressively": [205], "build": [206], "capabilities.": [208], "Through": [209], "experiments": [210], "mathematical": [212], "benchmarks": [214], "including": [215], "MATH,": [216], "AIME": [217], "2024,": [218], "TheoremQA,": [220], "demonstrate": [222], "our": [224, 293], "achieves": [228, 255], "comparable": [229], "distillation-based": [232], "10\u00d7": [236], "less": [237], "resources": [239], "no": [241], "best": [247], "7B": [248], "trained": [250], "solely": [251], "52.3%": [256], "MATH-500": [258], "(compared": [259], "54.1%": [261], "QwQ-distilled": [263], "baselines)": [264], "shows": [266], "emergent": [267], "self-correction": [268], "behaviors": [269], "23%": [271], "responses.": [273], "further": [275], "analyze": [276], "are": [280], "critical": [282], "provide": [288], "ablation": [289], "studies": [290], "demonstrating": [291], "essential": [298], "stable": [300], "training.": [301, 318], "All": [302], "code,": [303], "data,": [304], "will": [307], "be": [308], "released": [309], "facilitate": [311], "research": [313]}, "primary_topic": {"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.8090999722480774, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.8090999722480774, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10181", "display_name": "Natural Language Processing Techniques", "score": 0.04520000144839287, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T11307", "display_name": "Domain Adaptation and Few-Shot Learning", "score": 0.03060000017285347, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "doi:10.36227/techrxiv.176369747.70179230/v1", "is_oa": true, "landing_page_url": "https://doi.org/10.36227/techrxiv.176369747.70179230/v1", "pdf_url": "https://www.techrxiv.org/doi/pdf/10.36227/techrxiv.176369747.70179230/v1", "source": null, "license": "cc-by", "license_id": "https://openalex.org/licenses/cc-by", "version": "acceptedVersion", "is_accepted": true, "is_published": false, "raw_source_name": null, "raw_type": "posted-content"}, "cited_by_count": 0, "is_retracted": false}, {"id": "https://openalex.org/W4406734997", "doi": "https://doi.org/10.48550/arxiv.2501.11269", "title": "Can MLLMs Generalize to Multi-Party dialog? Exploring Multilingual Response Generation in Complex Scenarios", "publication_year": 2025, "publication_date": "2025-01-20", "type": "preprint", "language": "en", "authorships": [{"author_position": "first", "author": {"id": "https://openalex.org/A5044764986", "display_name": "Zhongtian Hu", "orcid": "https://orcid.org/0000-0001-5349-4751", "observed_orcids": ["https://orcid.org/0000-0001-5349-4751"]}, "institutions": [{"id": "https://openalex.org/I17145004", "display_name": "Northwestern Polytechnical University", "ror": "https://ror.org/01y0j0j86", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I17145004"]}, {"id": "https://openalex.org/I99418890", "display_name": "San Francisco Bay University", "ror": "https://ror.org/05wn69s11", "country_code": "US", "type": "education", "lineage": ["https://openalex.org/I99418890"]}], "countries": ["CN", "US"], "is_corresponding": false, "raw_author_name": "Hu, Zhongtian", "raw_affiliation_strings": ["School of Computer Science and Engineering , Northwestern Polytechnical University ,"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "School of Computer Science and Engineering , Northwestern Polytechnical University ,", "institution_ids": ["https://openalex.org/I17145004", "https://openalex.org/I99418890"]}]}, {"author_position": "middle", "author": {"id": null, "display_name": "Cui, Yiwen", "orcid": null, "observed_orcids": []}, "institutions": [{"id": "https://openalex.org/I17145004", "display_name": "Northwestern Polytechnical University", "ror": "https://ror.org/01y0j0j86", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I17145004"]}, {"id": "https://openalex.org/I99418890", "display_name": "San Francisco Bay University", "ror": "https://ror.org/05wn69s11", "country_code": "US", "type": "education", "lineage": ["https://openalex.org/I99418890"]}], "countries": ["CN", "US"], "is_corresponding": false, "raw_author_name": "Cui, Yiwen", "raw_affiliation_strings": ["School of Computer Science and Engineering , Northwestern Polytechnical University ,"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "School of Computer Science and Engineering , Northwestern Polytechnical University ,", "institution_ids": ["https://openalex.org/I17145004", "https://openalex.org/I99418890"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5059208730", "display_name": "Ronghan Li", "orcid": "https://orcid.org/0000-0002-1890-7845", "observed_orcids": ["https://orcid.org/0000-0002-1890-7845"]}, "institutions": [{"id": "https://openalex.org/I149594827", "display_name": "Xidian University", "ror": "https://ror.org/05s92vm98", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I149594827"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Li, Ronghan", "raw_affiliation_strings": ["School of Computer Science and Technology , Xidian University ,"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "School of Computer Science and Technology , Xidian University ,", "institution_ids": ["https://openalex.org/I149594827"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5066080136", "display_name": "Meng Zhao", "orcid": "https://orcid.org/0000-0003-4465-1553", "observed_orcids": ["https://orcid.org/0000-0003-4465-1553"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Zhao, Meng", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "last", "author": {"id": "https://openalex.org/A5100332311", "display_name": "Lifang Wang", "orcid": "https://orcid.org/0000-0003-4745-757X", "observed_orcids": ["https://orcid.org/0000-0003-4745-757X"]}, "institutions": [{"id": "https://openalex.org/I17145004", "display_name": "Northwestern Polytechnical University", "ror": "https://ror.org/01y0j0j86", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I17145004"]}, {"id": "https://openalex.org/I99418890", "display_name": "San Francisco Bay University", "ror": "https://ror.org/05wn69s11", "country_code": "US", "type": "education", "lineage": ["https://openalex.org/I99418890"]}], "countries": ["CN", "US"], "is_corresponding": false, "raw_author_name": "Wang, Lifang", "raw_affiliation_strings": ["School of Computer Science and Engineering , Northwestern Polytechnical University ,"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "School of Computer Science and Engineering , Northwestern Polytechnical University ,", "institution_ids": ["https://openalex.org/I17145004", "https://openalex.org/I99418890"]}]}], "abstract_inverted_index": {"Current": [0], "multilingual": [1, 23], "large": [2, 24], "language": [3], "models(MLLMs)": [4], "still": [5], "focus": [6], "on": [7, 55, 97, 133], "simple": [8], "question-answering": [9], "formats,": [10], "often": [11], "overlooking": [12], "more": [13, 47, 109], "complex": [14, 48], "dialogue": [15, 32, 99], "scenarios.": [16, 100], "In": [17], "other": [18], "words,": [19], "their": [20], "capabilities": [21], "of": [22, 115], "models": [25], "have": [26], "yet": [27], "to": [28, 46, 126, 128, 168], "be": [29], "validated": [30], "in": [31, 70, 103, 171], "tasks": [33], "with": [34, 138, 161], "intricate": [35], "structures.": [36], "We": [37], "therefore": [38], "ask,": [39], "Q1:": [40], "How": [41], "well": [42], "do": [43], "LLMs": [44], "generalize": [45, 127], "dialog": [49], "scenarios?": [50], "Q2:": [51], "Can": [52], "supervised": [53], "fine-tuning": [54], "a": [56, 80, 112, 145], "high-quality": [57, 81], "parallel": [58, 82, 94], "benchmark": [59], "restore": [60], "this": [61], "ability?": [62], "Q3:": [63], "Does": [64], "the": [65, 71, 92, 104, 139, 172], "\"multilingual": [66], "complementarity\"": [67], "effect": [68], "survive": [69], "setting?": [72], "To": [73], "answer": [74], "these": [75], "questions,": [76], "we": [77, 120], "introduce": [78], "XMP,": [79], "Multilingual": [83], "dataset": [84, 95, 105], "sourced": [85], "from": [86], "Multi-party": [87], "Podcast": [88], "dialogues,": [89], "which": [90], "is": [91, 158], "first": [93], "focusing": [96], "multi-party": [98, 129], "Most": [101], "samples": [102], "feature": [106], "three": [107], "or": [108], "participants,": [110], "discussing": [111], "wide": [113], "range": [114], "topics.": [116], "Through": [117], "extensive": [118], "experiments,": [119], "find": [121], "that,": [122], "R1:": [123], "MLLMs": [124], "fail": [125], "setting,": [130], "R2": [131], "Fine-tuning": [132], "XMP": [134], "improves": [135], "only": [136], "marginally,": [137], "70B": [140, 173], "model": [141], "achieving": [142], "at": [143], "most": [144], "1%": [146], "absolute": [147], "gain": [148], "over": [149], "its": [150], "8B": [151], "counterpart;": [152], "R3:": [153], "Mixing": [154], "languages": [155], "during": [156], "SFT": [157], "usually": [159], "detrimental,": [160], "any": [162], "benefits": [163], "being": [164], "marginal": [165], "and": [166], "limited": [167], "isolated": [169], "cases": [170], "model.": [174]}, "primary_topic": {"id": "https://openalex.org/T12031", "display_name": "Speech and dialogue systems", "score": 0.850600004196167, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T12031", "display_name": "Speech and dialogue systems", "score": 0.850600004196167, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.06870000064373016, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10181", "display_name": "Natural Language Processing Techniques", "score": 0.0625, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "pmh:oai:arXiv.org:2501.11269", "is_oa": true, "landing_page_url": "http://arxiv.org/abs/2501.11269", "pdf_url": "https://arxiv.org/pdf/2501.11269", "source": {"id": "https://openalex.org/S4306400194", "display_name": "arXiv (Cornell University)", "issn_l": "2331-8422", "issn": ["2331-8422"], "is_oa": true, "is_in_doaj": false, "is_core": false, "listed_in": [], "host_organization": "https://openalex.org/I205783295", "host_organization_name": "Cornell University", "host_organization_lineage": ["https://openalex.org/I205783295"], "host_organization_lineage_names": [], "type": "repository"}, "license": "cc-by", "license_id": "https://openalex.org/licenses/cc-by", "version": "submittedVersion", "is_accepted": false, "is_published": false, "raw_source_name": null, "raw_type": "text"}, "cited_by_count": 0, "is_retracted": false}, {"id": "https://openalex.org/W4415307344", "doi": "https://doi.org/10.48550/arxiv.2504.17432", "title": "Breaking the Modality Barrier: Universal Embedding Learning with Multimodal LLMs", "publication_year": 2025, "publication_date": "2025-04-24", "type": "preprint", "language": "en", "authorships": [{"author_position": "first", "author": {"id": "https://openalex.org/A5101333470", "display_name": "Tiancheng Gu", "orcid": "https://orcid.org/0009-0001-2302-246X", "observed_orcids": ["https://orcid.org/0009-0001-2302-246X"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Gu, Tiancheng", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5061907296", "display_name": "Kai-Cheng Yang", "orcid": "https://orcid.org/0000-0001-6381-4026", "observed_orcids": ["https://orcid.org/0000-0001-6381-4026"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Yang, Kaicheng", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5100531299", "display_name": "Ziyong Feng", "orcid": "https://orcid.org/0009-0007-8689-8366", "observed_orcids": ["https://orcid.org/0009-0007-8689-8366"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Feng, Ziyong", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5100740510", "display_name": "Xingjun Wang", "orcid": "https://orcid.org/0000-0001-5531-7273", "observed_orcids": ["https://orcid.org/0000-0001-5531-7273", "https://orcid.org/0000-0001-8206-2544"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Wang, Xingjun", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5103114439", "display_name": "Yanzhao Zhang", "orcid": "https://orcid.org/0000-0003-2894-4727", "observed_orcids": ["https://orcid.org/0000-0003-2894-4727", "https://orcid.org/0009-0003-6581-7783"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Zhang, Yanzhao", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5049484480", "display_name": "Dingkun Long", "orcid": "https://orcid.org/0000-0001-6570-9406", "observed_orcids": ["https://orcid.org/0000-0001-6570-9406"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Long, Dingkun", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5104162468", "display_name": "Yingda Chen", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Chen, Yingda", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5076697411", "display_name": "Weidong Cai", "orcid": "https://orcid.org/0000-0003-3706-8896", "observed_orcids": ["https://orcid.org/0000-0003-3706-8896"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Cai, Weidong", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "last", "author": {"id": "https://openalex.org/A5073110676", "display_name": "Jiankang Deng", "orcid": "https://orcid.org/0000-0002-3709-6216", "observed_orcids": ["https://orcid.org/0000-0002-3709-6216"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Deng, Jiankang", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}], "abstract_inverted_index": {"The": [0], "Contrastive": [1], "Language-Image": [2], "Pre-training": [3], "(CLIP)": [4], "framework": [5, 82], "has": [6], "become": [7], "a": [8, 79, 105], "widely": [9], "used": [10], "approach": [11, 164], "for": [12, 64, 90], "multimodal": [13, 67], "representation": [14, 135], "learning,": [15], "particularly": [16], "in": [17, 58, 175], "image-text": [18, 37], "retrieval": [19, 188, 195], "and": [20, 39, 144, 186, 192, 196, 213], "clustering.": [21], "However,": [22], "its": [23], "efficacy": [24], "is": [25], "constrained": [26], "by": [27], "three": [28], "key": [29], "limitations:": [30], "(1)": [31], "text": [32], "token": [33], "truncation,": [34], "(2)": [35], "isolated": [36], "encoding,": [38], "(3)": [40], "deficient": [41], "compositionality": [42], "due": [43], "to": [44, 86, 110, 131, 158], "bag-of-words": [45], "behavior.": [46], "While": [47], "recent": [48], "Multimodal": [49, 77], "Large": [50], "Language": [51], "Models": [52], "(MLLMs)": [53], "have": [54], "demonstrated": [55], "significant": [56], "advances": [57], "generalized": [59], "vision-language": [60], "understanding,": [61], "their": [62], "potential": [63], "learning": [65], "transferable": [66], "representations": [68, 89], "remains": [69], "underexplored.In": [70], "this": [71], "work,": [72], "we": [73, 98, 124, 138], "present": [74], "UniME": [75, 202], "(Universal": [76], "Embedding),": [78], "novel": [80], "two-stage": [81], "that": [83, 201], "leverages": [84], "MLLMs": [85], "learn": [87], "discriminative": [88, 101, 134, 168, 212], "diverse": [91], "downstream": [92, 176], "tasks.": [93, 177], "In": [94, 120], "the": [95, 112, 116, 121, 156, 183], "first": [96], "stage,": [97, 123], "perform": [99], "textual": [100], "knowledge": [102], "distillation": [103], "from": [104], "powerful": [106], "LLM-based": [107], "teacher": [108], "model": [109, 157], "enhance": [111], "embedding": [113], "capability": [114], "of": [115], "MLLM\u015b": [117], "language": [118], "component.": [119], "second": [122], "introduce": [125], "hard": [126, 148], "negative": [127, 142], "enhanced": [128], "instruction": [129], "tuning": [130], "further": [132], "advance": [133], "learning.": [136], "Specifically,": [137], "initially": [139], "mitigate": [140], "false": [141], "contamination": [143], "then": [145], "sample": [146], "multiple": [147, 187], "negatives": [149], "per": [150], "instance": [151], "within": [152], "each": [153], "batch,": [154], "forcing": [155], "focus": [159], "on": [160, 182], "challenging": [161], "samples.": [162], "This": [163], "not": [165], "only": [166], "improves": [167], "power": [169], "but": [170], "also": [171], "enhances": [172], "instruction-following": [173], "ability": [174], "We": [178], "conduct": [179], "extensive": [180], "experiments": [181], "MMEB": [184], "benchmark": [185], "tasks,": [189, 209], "including": [190], "short": [191], "long": [193], "caption": [194], "compositional": [197, 214], "retrieval.": [198], "Results": [199], "demonstrate": [200], "achieves": [203], "consistent": [204], "performance": [205], "improvement": [206], "across": [207], "all": [208], "exhibiting": [210], "superior": [211], "capabilities.": [215]}, "primary_topic": {"id": "https://openalex.org/T11714", "display_name": "Multimodal Machine Learning Applications", "score": 1.0, "subfield": {"id": "https://openalex.org/subfields/1707", "display_name": "Computer Vision and Pattern Recognition"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T11714", "display_name": "Multimodal Machine Learning Applications", "score": 1.0, "subfield": {"id": "https://openalex.org/subfields/1707", "display_name": "Computer Vision and Pattern Recognition"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10627", "display_name": "Advanced Image and Video Retrieval Techniques", "score": 0.0, "subfield": {"id": "https://openalex.org/subfields/1707", "display_name": "Computer Vision and Pattern Recognition"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.0, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "pmh:oai:arXiv.org:2504.17432", "is_oa": true, "landing_page_url": "http://arxiv.org/abs/2504.17432", "pdf_url": "https://arxiv.org/pdf/2504.17432", "source": {"id": "https://openalex.org/S4306400194", "display_name": "arXiv (Cornell University)", "issn_l": "2331-8422", "issn": ["2331-8422"], "is_oa": true, "is_in_doaj": false, "is_core": false, "listed_in": [], "host_organization": "https://openalex.org/I205783295", "host_organization_name": "Cornell University", "host_organization_lineage": ["https://openalex.org/I205783295"], "host_organization_lineage_names": [], "type": "repository"}, "license": "cc-by", "license_id": "https://openalex.org/licenses/cc-by", "version": "submittedVersion", "is_accepted": false, "is_published": false, "raw_source_name": null, "raw_type": "text"}, "cited_by_count": 0, "is_retracted": false}, {"id": "https://openalex.org/W4414892782", "doi": "https://doi.org/10.48550/arxiv.2506.00743", "title": "Assortment of Attention Heads: Accelerating Federated PEFT with Head Pruning and Strategic Client Selection", "publication_year": 2025, "publication_date": "2025-05-31", "type": "preprint", "language": "en", "authorships": [{"author_position": "first", "author": {"id": "https://openalex.org/A5079036687", "display_name": "Yeshwanth Venkatesha", "orcid": "https://orcid.org/0000-0001-8406-9635", "observed_orcids": ["https://orcid.org/0000-0001-8406-9635"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Venkatesha, Yeshwanth", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5087095284", "display_name": "Souvik Kundu", "orcid": "https://orcid.org/0000-0001-5815-8765", "observed_orcids": ["https://orcid.org/0000-0001-5815-8765"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Kundu, Souvik", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "last", "author": {"id": "https://openalex.org/A5050310538", "display_name": "Priyadarshini Panda", "orcid": "https://orcid.org/0000-0002-4167-6782", "observed_orcids": ["https://orcid.org/0000-0002-4167-6782"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Panda, Priyadarshini", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}], "abstract_inverted_index": {"Parameter": [0], "Efficient": [1], "Fine-Tuning": [2], "(PEFT)": [3], "has": [4], "become": [5], "the": [6, 69, 81, 103, 107, 113, 116, 124, 142, 156, 194], "de-facto": [7], "approach": [8], "in": [9, 18, 25, 176, 187], "adapting": [10], "Large": [11], "Language": [12, 20], "Models": [13], "(LLMs)": [14], "for": [15, 72], "downstream": [16], "tasks": [17], "Natural": [19], "Processing.": [21], "However,": [22], "its": [23], "adoption": [24], "privacy-preserving": [26], "distributed": [27], "learning": [28], "frameworks,": [29], "such": [30, 47], "as": [31, 48, 164], "Federated": [32], "Learning": [33], "(FL),": [34], "remains": [35], "relatively": [36], "limited.": [37], "This": [38], "is": [39], "mainly": [40], "due": [41], "to": [42, 45, 65, 173, 182], "challenges": [43, 82], "specific": [44], "FL,": [46], "resource-constrained": [49], "devices": [50], "and": [51, 92, 150, 159, 184], "diverse": [52, 131], "data": [53], "distributions": [54], "among": [55], "clients.": [56], "In": [57], "this": [58], "paper,": [59], "we": [60], "propose": [61], "an": [62], "efficient": [63], "method": [64], "perform": [66], "PEFT": [67, 166], "within": [68, 102], "FL": [70], "framework": [71], "Multi-Head": [73], "Attention": [74], "(MHA)": [75], "based": [76, 111], "language": [77], "models.": [78], "We": [79, 138, 154], "address": [80], "through": [83], "head": [84], "pruning,": [85], "a": [86, 93, 177, 185], "novel": [87], "head-specific": [88], "weighted": [89], "aggregation": [90, 120], "mechanism,": [91], "client": [94, 135], "selection": [95, 136], "strategy.": [96, 137], "Head": [97], "pruning": [98], "minimizes": [99], "training": [100, 188], "complexity": [101], "clients,": [104], "guided": [105], "by": [106], "importance": [108], "score": [109], "computed": [110], "on": [112, 141], "confidence": [114], "of": [115, 121, 171, 180, 190], "attention": [117], "head.": [118], "Weighted": [119], "heads": [122], "ensures": [123], "global": [125], "model": [126, 161], "captures": [127], "crucial": [128], "updates": [129], "from": [130], "clients": [132], "complementing": [133], "our": [134, 165], "show": [139], "results": [140], "MultiNLI": [143, 157], "benchmark": [144], "along": [145], "with": [146, 162], "20": [147], "Newsgroups,": [148], "XL-Sum,": [149], "E2E": [151], "NLG": [152], "datasets.": [153], "use": [155], "dataset": [158], "T5-small": [160], "LoRA": [163], "method,": [167], "attaining": [168], "sparsity": [169], "levels": [170], "up": [172, 181], "90%,": [174], "resulting": [175], "communication": [178], "advantage": [179], "1.8x": [183], "reduction": [186], "OPs": [189], "3.9x": [191], "while": [192], "maintaining": [193], "accuracy": [195], "drop": [196], "under": [197], "2%.": [198]}, "primary_topic": {"id": "https://openalex.org/T10764", "display_name": "Privacy-Preserving Technologies in Data", "score": 0.9961000084877014, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T10764", "display_name": "Privacy-Preserving Technologies in Data", "score": 0.9961000084877014, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T11307", "display_name": "Domain Adaptation and Few-Shot Learning", "score": 0.0017999999690800905, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.0006000000284984708, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "pmh:oai:arXiv.org:2506.00743", "is_oa": true, "landing_page_url": "http://arxiv.org/abs/2506.00743", "pdf_url": "https://arxiv.org/pdf/2506.00743", "source": {"id": "https://openalex.org/S4306400194", "display_name": "arXiv (Cornell University)", "issn_l": "2331-8422", "issn": ["2331-8422"], "is_oa": true, "is_in_doaj": false, "is_core": false, "listed_in": [], "host_organization": "https://openalex.org/I205783295", "host_organization_name": "Cornell University", "host_organization_lineage": ["https://openalex.org/I205783295"], "host_organization_lineage_names": [], "type": "repository"}, "license": null, "license_id": null, "version": "submittedVersion", "is_accepted": false, "is_published": false, "raw_source_name": null, "raw_type": "text"}, "cited_by_count": 0, "is_retracted": false}, {"id": "https://openalex.org/W4416157480", "doi": "https://doi.org/10.48550/arxiv.2511.06360", "title": "AesTest: Measuring Aesthetic Intelligence from Perception to Production", "publication_year": 2025, "publication_date": "2025-11-09", "type": "preprint", "language": null, "authorships": [{"author_position": "first", "author": {"id": null, "display_name": "Wang, Guolong", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Wang, Guolong", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5060016795", "display_name": "Heng Huang", "orcid": "https://orcid.org/0000-0002-3483-8333", "observed_orcids": ["https://orcid.org/0000-0002-3483-8333"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Huang, Heng", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5100346716", "display_name": "Zhiqiang Zhang", "orcid": "https://orcid.org/0000-0002-7657-0435", "observed_orcids": ["https://orcid.org/0000-0002-7657-0435"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Zhang, Zhiqiang", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5057425369", "display_name": "Wentian Li", "orcid": "https://orcid.org/0000-0003-1155-110X", "observed_orcids": ["https://orcid.org/0000-0003-1155-110X", "https://orcid.org/0009-0003-8496-6776"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Li, Wentian", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5028668424", "display_name": "Ma Feilong", "orcid": "https://orcid.org/0000-0002-6838-3971", "observed_orcids": ["https://orcid.org/0000-0002-6838-3971"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Ma, Feilong", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "last", "author": {"id": "https://openalex.org/A5100641352", "display_name": "Xin Jin", "orcid": "https://orcid.org/0000-0003-3873-1653", "observed_orcids": ["https://orcid.org/0000-0003-3873-1653"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Jin, Xin", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}], "abstract_inverted_index": {"Perceiving": [0], "and": [1, 54, 75, 101, 111, 129, 138], "producing": [2], "aesthetic": [3, 22, 38, 52, 118, 148], "judgments": [4], "is": [5], "a": [6, 47], "fundamental": [7], "yet": [8], "underexplored": [9], "capability": [10], "for": [11, 20, 50], "multimodal": [12, 51], "large": [13], "language": [14], "models": [15], "(MLLMs).": [16], "However,": [17], "existing": [18], "benchmarks": [19], "image": [21], "assessment": [23], "(IAA)": [24], "are": [25, 79], "narrow": [26], "in": [27, 81, 146, 159], "perception": [28, 53], "scope": [29], "or": [30], "lack": [31], "the": [32, 58], "diversity": [33], "needed": [34], "to": [35, 155], "evaluate": [36, 133], "systematic": [37], "production.": [39], "To": [40], "address": [41], "this": [42, 160], "gap,": [43], "we": [44], "introduce": [45], "AesTest,": [46, 142], "comprehensive": [48], "benchmark": [49], "production,": [55], "distinguished": [56], "by": [57], "following": [59], "features:": [60], "1)": [61], "It": [62, 88, 104, 115], "consists": [63], "of": [64, 84, 107], "curated": [65], "multiple-choice": [66], "questions": [67], "spanning": [68], "ten": [69], "tasks,": [70], "covering": [71], "perception,": [72], "appreciation,": [73], "creation,": [74], "photography.": [76], "These": [77], "tasks": [78], "grounded": [80], "psychological": [82], "theories": [83], "generative": [85], "learning.": [86], "2)": [87], "integrates": [89], "data": [90], "from": [91], "diverse": [92], "sources,": [93], "including": [94], "professional": [95], "editing": [96], "workflows,": [97], "photographic": [98], "composition": [99], "tutorials,": [100], "crowdsourced": [102], "preferences.": [103], "ensures": [105], "coverage": [106], "both": [108, 134], "expert-level": [109], "principles": [110], "real-world": [112], "variation.": [113], "3)": [114], "supports": [116], "various": [117], "query": [119], "types,": [120], "such": [121], "as": [122], "attribute-based": [123], "analysis,": [124], "emotional": [125], "resonance,": [126], "compositional": [127], "choice,": [128], "stylistic": [130], "reasoning.": [131], "We": [132, 150], "instruction-tuned": [135], "IAA": [136], "MLLMs": [137, 140], "general": [139], "on": [141], "revealing": [143], "significant": [144], "challenges": [145], "building": [147], "intelligence.": [149], "will": [151], "publicly": [152], "release": [153], "AesTest": [154], "support": [156], "future": [157], "research": [158], "area.": [161]}, "primary_topic": {"id": "https://openalex.org/T11714", "display_name": "Multimodal Machine Learning Applications", "score": 0.8715999722480774, "subfield": {"id": "https://openalex.org/subfields/1707", "display_name": "Computer Vision and Pattern Recognition"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T11714", "display_name": "Multimodal Machine Learning Applications", "score": 0.8715999722480774, "subfield": {"id": "https://openalex.org/subfields/1707", "display_name": "Computer Vision and Pattern Recognition"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T12650", "display_name": "Aesthetic Perception and Analysis", "score": 0.06509999930858612, "subfield": {"id": "https://openalex.org/subfields/2805", "display_name": "Cognitive Neuroscience"}, "field": {"id": "https://openalex.org/fields/28", "display_name": "Neuroscience"}, "domain": {"id": "https://openalex.org/domains/1", "display_name": "Life Sciences"}}, {"id": "https://openalex.org/T10775", "display_name": "Generative Adversarial Networks and Image Synthesis", "score": 0.025499999523162842, "subfield": {"id": "https://openalex.org/subfields/1707", "display_name": "Computer Vision and Pattern Recognition"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "pmh:oai:arXiv.org:2511.06360", "is_oa": true, "landing_page_url": "http://arxiv.org/abs/2511.06360", "pdf_url": "https://arxiv.org/pdf/2511.06360", "source": {"id": "https://openalex.org/S4306400194", "display_name": "arXiv (Cornell University)", "issn_l": "2331-8422", "issn": ["2331-8422"], "is_oa": true, "is_in_doaj": false, "is_core": false, "listed_in": [], "host_organization": "https://openalex.org/I205783295", "host_organization_name": "Cornell University", "host_organization_lineage": ["https://openalex.org/I205783295"], "host_organization_lineage_names": [], "type": "repository"}, "license": null, "license_id": null, "version": "submittedVersion", "is_accepted": false, "is_published": false, "raw_source_name": null, "raw_type": "text"}, "cited_by_count": 0, "is_retracted": false}, {"id": "https://openalex.org/W4416051146", "doi": "https://doi.org/10.48550/arxiv.2508.15471", "title": "SLM4Offer: Personalized Marketing Offer Generation Using Contrastive Learning Based Fine-Tuning", "publication_year": 2025, "publication_date": "2025-08-21", "type": "preprint", "language": "en", "authorships": [{"author_position": "first", "author": {"id": "https://openalex.org/A5120318607", "display_name": "Vedasamhitha Challapalli", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Challapalli, Vedasamhitha", "raw_affiliation_strings": ["SPARC Research , HCLSoftware , Bengaluru"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "SPARC Research , HCLSoftware , Bengaluru", "institution_ids": []}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5119715234", "display_name": "Konduru Venkat Sai", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Sai, Konduru Venkat", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5023905510", "display_name": "Piyush Pratap Singh", "orcid": "https://orcid.org/0000-0002-6346-000X", "observed_orcids": ["https://orcid.org/0000-0002-6346-000X"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Singh, Piyush Pratap", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": null, "display_name": "Prasad, Rupesh", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Prasad, Rupesh", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": null, "display_name": "Maurya, Arvind", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Maurya, Arvind", "raw_affiliation_strings": ["SPARC Research , HCLSoftware , Bengaluru"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "SPARC Research , HCLSoftware , Bengaluru", "institution_ids": []}]}, {"author_position": "last", "author": {"id": "https://openalex.org/A5027090470", "display_name": "Atul Kumar Singh", "orcid": "https://orcid.org/0000-0003-0081-6109", "observed_orcids": ["https://orcid.org/0000-0003-0081-6109"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Singh, Atul", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}], "abstract_inverted_index": {"Personalized": [0], "marketing": [1], "has": [2], "emerged": [3], "as": [4], "a": [5, 76, 87, 100, 119, 157, 172, 181], "pivotal": [6], "strategy": [7], "for": [8, 37, 69, 80], "enhancing": [9], "customer": [10, 43, 113, 163], "engagement": [11], "and": [12, 17, 26, 41, 145, 154, 165], "driving": [13], "business": [14], "growth.": [15], "Academic": [16], "industry": [18], "efforts": [19], "have": [20], "predominantly": [21], "focused": [22], "on": [23, 156], "recommendation": [24], "systems": [25], "personalized": [27, 81, 193], "advertisements.": [28], "Nonetheless,": [29], "this": [30], "facet": [31], "of": [32, 64, 188], "personalization": [33, 50], "holds": [34], "significant": [35], "potential": [36], "increasing": [38], "conversion": [39], "rates": [40], "improving": [42], "satisfaction.": [44], "Prior": [45], "studies": [46], "suggest": [47], "that": [48], "well-executed": [49], "strategies": [51], "can": [52], "boost": [53], "revenue": [54], "by": [55, 85, 135], "up": [56], "to": [57, 111, 161], "40": [58], "percent,": [59], "underscoring": [60], "the": [61, 130, 140, 147, 186], "strategic": [62], "importance": [63], "developing": [65], "intelligent,": [66], "data-driven": [67], "approaches": [68], "offer": [70, 82, 166, 177], "generation.": [71], "This": [72], "work": [73], "introduces": [74], "SLM4Offer,": [75], "generative": [77], "AI": [78], "model": [79, 151], "generation,": [83], "developed": [84], "fine-tuning": [86, 183], "pre-trained": [88], "encoder-decoder": [89], "language": [90], "model,": [91], "specifically": [92], "Google's": [93], "Text-to-Text": [94], "Transfer": [95], "Transformer": [96], "(T5-Small": [97], "60M)": [98], "using": [99], "contrastive": [101, 136, 189], "learning": [102, 132], "approach.": [103], "SLM4Offer": [104, 127], "employs": [105], "InfoNCE": [106], "(Information": [107], "Noise-Contrastive": [108], "Estimation)": [109], "loss": [110], "align": [112], "personas": [114], "with": [115], "relevant": [116], "offers": [117], "in": [118, 126, 129, 176, 191], "shared": [120], "embedding": [121], "space.": [122], "A": [123], "key": [124], "innovation": [125], "lies": [128], "adaptive": [131], "behaviour": [133, 164], "introduced": [134], "loss,": [137], "which": [138], "reshapes": [139], "latent": [141], "space": [142], "during": [143], "training": [144], "enhances": [146], "model's": [148], "generalizability.": [149], "The": [150], "is": [152], "fine-tuned": [153], "evaluated": [155], "synthetic": [158], "dataset": [159], "designed": [160], "simulate": [162], "acceptance": [167, 178], "patterns.": [168], "Experimental": [169], "results": [170], "demonstrate": [171], "17": [173], "percent": [174], "improvement": [175], "rate": [179], "over": [180], "supervised": [182], "baseline,": [184], "highlighting": [185], "effectiveness": [187], "objectives": [190], "advancing": [192], "marketing.": [194]}, "primary_topic": {"id": "https://openalex.org/T10203", "display_name": "Recommender Systems and Techniques", "score": 0.6606000065803528, "subfield": {"id": "https://openalex.org/subfields/1710", "display_name": "Information Systems"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T10203", "display_name": "Recommender Systems and Techniques", "score": 0.6606000065803528, "subfield": {"id": "https://openalex.org/subfields/1710", "display_name": "Information Systems"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T12384", "display_name": "Customer churn and segmentation", "score": 0.1111999973654747, "subfield": {"id": "https://openalex.org/subfields/1406", "display_name": "Marketing"}, "field": {"id": "https://openalex.org/fields/14", "display_name": "Business, Management and Accounting"}, "domain": {"id": "https://openalex.org/domains/2", "display_name": "Social Sciences"}}, {"id": "https://openalex.org/T10609", "display_name": "Digital Marketing and Social Media", "score": 0.04540000110864639, "subfield": {"id": "https://openalex.org/subfields/3312", "display_name": "Sociology and Political Science"}, "field": {"id": "https://openalex.org/fields/33", "display_name": "Social Sciences"}, "domain": {"id": "https://openalex.org/domains/2", "display_name": "Social Sciences"}}], "primary_location": {"id": "pmh:oai:arXiv.org:2508.15471", "is_oa": true, "landing_page_url": "http://arxiv.org/abs/2508.15471", "pdf_url": "https://arxiv.org/pdf/2508.15471", "source": {"id": "https://openalex.org/S4306400194", "display_name": "arXiv (Cornell University)", "issn_l": "2331-8422", "issn": ["2331-8422"], "is_oa": true, "is_in_doaj": false, "is_core": false, "listed_in": [], "host_organization": "https://openalex.org/I205783295", "host_organization_name": "Cornell University", "host_organization_lineage": ["https://openalex.org/I205783295"], "host_organization_lineage_names": [], "type": "repository"}, "license": null, "license_id": null, "version": "submittedVersion", "is_accepted": false, "is_published": false, "raw_source_name": null, "raw_type": "text"}, "cited_by_count": 0, "is_retracted": false}, {"id": "https://openalex.org/W7114985404", "doi": "https://doi.org/10.24405/21759", "title": "Beyond statistical parroting: hard-coding truth into LLMs via ontologies", "publication_year": 2025, "publication_date": "2025-10-27", "type": "article", "language": "en", "authorships": [{"author_position": "first", "author": {"id": null, "display_name": "Giglou, Hamed Babaei", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Giglou, Hamed Babaei", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5119949186", "display_name": "Simon Burbach", "orcid": "https://orcid.org/0009-0009-8373-6622", "observed_orcids": ["https://orcid.org/0009-0009-8373-6622"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Burbach, Simon", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": null, "display_name": "Compagno, Francesco", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Compagno, Francesco", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": null, "display_name": "Ismail, Muhammad", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Ismail, Muhammad", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": null, "display_name": "Singh, Gunjan", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Singh, Gunjan", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "last", "author": {"id": null, "display_name": "Rudolph, Sebastian", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Rudolph, Sebastian", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}], "abstract_inverted_index": {"Large": [0], "Language": [1], "Models": [2], "(LLMs)": [3], "are": [4], "powerful": [5], "but": [6], "prone": [7], "to": [8, 34, 58, 82], "hallucinations": [9], "and": [10, 38, 76, 131], "factual": [11, 36, 62], "inconsistencies,": [12], "especially": [13], "in": [14], "knowledge-intensive": [15], "tasks.": [16], "In": [17], "this": [18], "work,": [19], "we": [20, 47], "explore": [21], "the": [22, 41, 61, 74, 123, 133], "integration": [23], "of": [24, 52, 64, 89, 96], "structured": [25], "ontological": [26, 103], "knowledge": [27, 114], "into": [28, 105], "LLM": [29, 65], "prompts": [30, 57, 106], "as": [31, 44], "a": [32, 45, 87], "strategy": [33], "enhance": [35], "accuracy": [37, 63], "reliability.": [39], "Using": [40], "Pizza": [42], "Ontology": [43], "showcase,": [46], "probe": [48], "how": [49], "different": [50, 69], "levels": [51], "domain": [53, 113], "grounding\u2014ranging": [54], "from": [55, 73, 80], "base": [56], "ontology-informed": [59], "prompts\u2014affect": [60], "responses.": [66], "We": [67], "tested": [68], "instruction-tuned": [70], "last-generation": [71], "LLMs": [72, 130], "Qwen": [75], "Llama": [77], "families,": [78], "ranging": [79], "0.5": [81], "72": [83], "billion": [84], "parameters,": [85], "on": [86], "dataset": [88], "approximately": [90], "51": [91], "questions": [92], "requiring": [93], "various": [94], "types": [95], "reasoning.": [97], "Our": [98, 141], "results": [99], "show": [100], "that": [101, 111], "injecting": [102], "axioms": [104], "improves": [107], "response": [108], "accuracy,": [109], "demonstrating": [110], "formal": [112], "can": [115], "significantly": [116], "reduce": [117], "hallucinations.": [118], "This": [119], "proof-of-concept": [120], "study": [121], "highlights": [122], "potential": [124], "for": [125, 135], "combining": [126], "symbolic": [127], "approaches": [128], "with": [129], "lays": [132], "groundwork": [134], "more": [136], "reliable,": [137], "explainable": [138], "AI": [139], "systems.": [140], "codebase": [142], "is": [143], "available": [144], "at": [145], "https://github.com/HamedBabaei/OntoTruth.": [146]}, "primary_topic": {"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.8285999894142151, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.8285999894142151, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10215", "display_name": "Semantic Web and Ontologies", "score": 0.10580000281333923, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10181", "display_name": "Natural Language Processing Techniques", "score": 0.020400000736117363, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "doi:10.24405/21759", "is_oa": true, "landing_page_url": "https://doi.org/10.24405/21759", "pdf_url": null, "source": {"id": "https://openalex.org/S7407052928", "display_name": "OpenHSU", "issn_l": null, "issn": null, "is_oa": false, "is_in_doaj": false, "is_core": false, "listed_in": [], "host_organization": null, "host_organization_name": null, "host_organization_lineage": [], "host_organization_lineage_names": [], "type": "repository"}, "license": null, "license_id": null, "version": null, "is_accepted": false, "is_published": false, "raw_source_name": null, "raw_type": "article-journal"}, "cited_by_count": 0, "is_retracted": false}, {"id": "https://openalex.org/W4415101324", "doi": "https://doi.org/10.48550/arxiv.2503.09025", "title": "Aligning to What? Limits to RLHF Based Alignment", "publication_year": 2025, "publication_date": "2025-03-12", "type": "preprint", "language": "en", "authorships": [{"author_position": "first", "author": {"id": "https://openalex.org/A5117865201", "display_name": "Logan Barnhart", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Barnhart, Logan", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5092675602", "display_name": "Reza Akbarian Bafghi", "orcid": "https://orcid.org/0009-0006-7372-2516", "observed_orcids": ["https://orcid.org/0009-0006-7372-2516"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Bafghi, Reza Akbarian", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5086042756", "display_name": "Stephen Becker", "orcid": "https://orcid.org/0000-0002-1932-8159", "observed_orcids": ["https://orcid.org/0000-0002-1932-8159"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Becker, Stephen", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "last", "author": {"id": "https://openalex.org/A5012536010", "display_name": "Maziar Raissi", "orcid": "https://orcid.org/0000-0002-8467-4568", "observed_orcids": ["https://orcid.org/0000-0002-8467-4568"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Raissi, Maziar", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}], "abstract_inverted_index": {"Reinforcement": [0], "Learning": [1], "from": [2], "Human": [3], "Feedback": [4], "(RLHF)": [5], "is": [6], "increasingly": [7], "used": [8], "to": [9, 60, 114], "align": [10], "large": [11], "language": [12], "models": [13, 74, 91], "(LLMs)": [14], "with": [15, 86], "human": [16], "preferences.": [17], "However,": [18], "the": [19, 32, 66, 72, 109, 140], "effectiveness": [20], "of": [21, 71], "RLHF": [22, 35, 54, 102], "in": [23, 42], "addressing": [24], "underlying": [25], "biases": [26, 41, 47, 70, 113], "remains": [27], "unclear.": [28], "This": [29], "study": [30], "investigates": [31], "relationship": [33], "between": [34], "and": [36, 39, 58, 64, 68, 78, 92], "both": [37], "covert": [38, 67, 137], "overt": [40, 69], "LLMs,": [43], "particularly": [44], "focusing": [45], "on": [46, 88], "against": [48], "African": [49], "Americans.": [50], "We": [51, 82], "applied": [52], "various": [53], "techniques": [55, 128], "(DPO,": [56], "ORPO,": [57], "RLOO)": [59], "Llama": [61], "3": [62], "8B": [63], "evaluated": [65], "resulting": [73], "using": [75], "matched-guise": [76], "probing": [77], "explicit": [79], "bias": [80], "testing.": [81], "performed": [83], "additional": [84], "tests": [85], "DPO": [87], "different": [89], "base": [90], "datasets;": [93], "among": [94], "several": [95], "implications,": [96], "we": [97, 107, 120], "found": [98], "that": [99, 123, 125], "SFT": [100], "before": [101], "calcifies": [103], "model": [104], "biases.": [105], "Additionally,": [106], "extend": [108], "tools": [110], "for": [111, 131, 142], "measuring": [112], "multi-modal": [115], "models.": [116], "Through": [117], "our": [118], "experiments": [119], "collect": [121], "evidence": [122], "indicates": [124], "current": [126], "alignment": [127, 149], "are": [129], "inadequate": [130], "nebulous": [132], "tasks": [133], "such": [134], "as": [135], "mitigating": [136], "biases,": [138], "highlighting": [139], "need": [141], "capable": [143], "datasets,": [144], "data": [145], "curating": [146], "techniques,": [147], "or": [148], "tools.": [150]}, "primary_topic": {"id": "https://openalex.org/T10883", "display_name": "Ethics and Social Impacts of AI", "score": 0.9419000148773193, "subfield": {"id": "https://openalex.org/subfields/3311", "display_name": "Safety Research"}, "field": {"id": "https://openalex.org/fields/33", "display_name": "Social Sciences"}, "domain": {"id": "https://openalex.org/domains/2", "display_name": "Social Sciences"}}, "topics": [{"id": "https://openalex.org/T10883", "display_name": "Ethics and Social Impacts of AI", "score": 0.9419000148773193, "subfield": {"id": "https://openalex.org/subfields/3311", "display_name": "Safety Research"}, "field": {"id": "https://openalex.org/fields/33", "display_name": "Social Sciences"}, "domain": {"id": "https://openalex.org/domains/2", "display_name": "Social Sciences"}}, {"id": "https://openalex.org/T12262", "display_name": "Hate Speech and Cyberbullying Detection", "score": 0.010599999688565731, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.008700000122189522, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "pmh:oai:arXiv.org:2503.09025", "is_oa": true, "landing_page_url": "http://arxiv.org/abs/2503.09025", "pdf_url": "https://arxiv.org/pdf/2503.09025", "source": {"id": "https://openalex.org/S4306400194", "display_name": "arXiv (Cornell University)", "issn_l": "2331-8422", "issn": ["2331-8422"], "is_oa": true, "is_in_doaj": false, "is_core": false, "listed_in": [], "host_organization": "https://openalex.org/I205783295", "host_organization_name": "Cornell University", "host_organization_lineage": ["https://openalex.org/I205783295"], "host_organization_lineage_names": [], "type": "repository"}, "license": null, "license_id": null, "version": "submittedVersion", "is_accepted": false, "is_published": false, "raw_source_name": null, "raw_type": "text"}, "cited_by_count": 0, "is_retracted": false}, {"id": "https://openalex.org/W4417183029", "doi": "https://doi.org/10.48550/arxiv.2507.01492", "title": "AVC-DPO: Aligned Video Captioning via Direct Preference Optimization", "publication_year": 2025, "publication_date": "2025-07-02", "type": "preprint", "language": "en", "authorships": [{"author_position": "first", "author": {"id": "https://openalex.org/A5101730616", "display_name": "Jiyang Tang", "orcid": "https://orcid.org/0009-0008-3229-7976", "observed_orcids": ["https://orcid.org/0009-0008-3229-7976"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Tang, Jiyang", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5032417356", "display_name": "Hengyi Li", "orcid": "https://orcid.org/0000-0003-4112-7297", "observed_orcids": ["https://orcid.org/0000-0003-4112-7297"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Li, Hengyi", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5108328077", "display_name": "Yifan Du", "orcid": "https://orcid.org/0009-0002-4042-525X", "observed_orcids": ["https://orcid.org/0009-0002-4042-525X"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Du, Yifan", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "last", "author": {"id": "https://openalex.org/A5037145565", "display_name": "Wayne Xin Zhao", "orcid": "https://orcid.org/0000-0002-8333-6196", "observed_orcids": ["https://orcid.org/0000-0002-8333-6196"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Zhao, Wayne Xin", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}], "abstract_inverted_index": {"Although": [0], "video": [1, 13, 25, 54], "multimodal": [2], "large": [3], "language": [4], "models": [5], "(video": [6], "MLLMs)": [7], "have": [8, 109], "achieved": [9, 110], "substantial": [10], "progress": [11], "in": [12, 53, 113], "captioning": [14, 51], "tasks,": [15], "it": [16], "remains": [17], "challenging": [18], "to": [19, 28, 49, 98, 134], "adjust": [20], "the": [21, 87, 114, 127, 135], "focal": [22], "emphasis": [23], "of": [24], "captions": [26], "according": [27, 133], "human": [29], "preferences.": [30, 84], "To": [31], "address": [32], "this": [33, 106], "limitation,": [34], "we": [35, 108], "propose": [36], "Aligned": [37], "Video": [38, 119, 128], "Captioning": [39, 121, 130], "via": [40], "Direct": [41], "Preference": [42], "Optimization": [43], "(AVC-DPO),": [44], "a": [45, 80], "post-training": [46], "framework": [47], "designed": [48], "enhance": [50], "capabilities": [52], "MLLMs": [55], "through": [56], "preference": [57], "alignment.": [58, 104], "Our": [59], "approach": [60], "designs": [61], "enhanced": [62], "prompts": [63], "that": [64, 74], "specifically": [65], "target": [66], "temporal": [67], "dynamics": [68], "and": [69, 102], "spatial": [70], "information-two": [71], "key": [72], "factors": [73], "humans": [75], "care": [76], "about": [77], "when": [78], "watching": [79], "video-thereby": [81], "incorporating": [82], "human-centric": [83], "AVC-DPO": [85], "leverages": [86], "same": [88], "foundation": [89], "model's": [90], "caption": [91, 103], "generation": [92], "responses": [93], "under": [94], "varied": [95], "prompt": [96], "conditions": [97], "conduct": [99], "preference-aware": [100], "training": [101], "Using": [105], "framework,": [107], "exceptional": [111], "performance": [112], "LOVE@CVPR'25": [115], "Workshop": [116], "Track": [117], "1A:": [118], "Detailed": [120, 129], "Challenge,": [122], "achieving": [123], "first": [124], "place": [125], "on": [126], "(VDC)": [131], "benchmark": [132], "VDCSCORE": [136], "evaluation": [137], "metric.": [138]}, "primary_topic": {"id": "https://openalex.org/T11714", "display_name": "Multimodal Machine Learning Applications", "score": 1.0, "subfield": {"id": "https://openalex.org/subfields/1707", "display_name": "Computer Vision and Pattern Recognition"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T11714", "display_name": "Multimodal Machine Learning Applications", "score": 1.0, "subfield": {"id": "https://openalex.org/subfields/1707", "display_name": "Computer Vision and Pattern Recognition"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10775", "display_name": "Generative Adversarial Networks and Image Synthesis", "score": 0.0, "subfield": {"id": "https://openalex.org/subfields/1707", "display_name": "Computer Vision and Pattern Recognition"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10812", "display_name": "Human Pose and Action Recognition", "score": 0.0, "subfield": {"id": "https://openalex.org/subfields/1707", "display_name": "Computer Vision and Pattern Recognition"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "pmh:oai:arXiv.org:2507.01492", "is_oa": true, "landing_page_url": "http://arxiv.org/abs/2507.01492", "pdf_url": "https://arxiv.org/pdf/2507.01492", "source": {"id": "https://openalex.org/S4306400194", "display_name": "arXiv (Cornell University)", "issn_l": "2331-8422", "issn": ["2331-8422"], "is_oa": true, "is_in_doaj": false, "is_core": false, "listed_in": [], "host_organization": "https://openalex.org/I205783295", "host_organization_name": "Cornell University", "host_organization_lineage": ["https://openalex.org/I205783295"], "host_organization_lineage_names": [], "type": "repository"}, "license": null, "license_id": null, "version": "submittedVersion", "is_accepted": false, "is_published": false, "raw_source_name": null, "raw_type": "text"}, "cited_by_count": 0, "is_retracted": false}, {"id": "https://openalex.org/W4414689509", "doi": "https://doi.org/10.48550/arxiv.2506.02015", "title": "OSPO: Object-Centric Self-Improving Preference Optimization for Text-to-Image Generation", "publication_year": 2025, "publication_date": "2025-05-28", "type": "preprint", "language": "en", "authorships": [{"author_position": "first", "author": {"id": null, "display_name": "Oh, Yoonjin", "orcid": null, "observed_orcids": []}, "institutions": [{"id": "https://openalex.org/I197347611", "display_name": "Korea University", "ror": "https://ror.org/047dqcg40", "country_code": "KR", "type": "education", "lineage": ["https://openalex.org/I197347611"]}], "countries": ["KR"], "is_corresponding": false, "raw_author_name": "Oh, Yoonjin", "raw_affiliation_strings": ["Korea University"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Korea University", "institution_ids": ["https://openalex.org/I197347611"]}]}, {"author_position": "middle", "author": {"id": null, "display_name": "Kim, Yongjin", "orcid": null, "observed_orcids": []}, "institutions": [{"id": "https://openalex.org/I197347611", "display_name": "Korea University", "ror": "https://ror.org/047dqcg40", "country_code": "KR", "type": "education", "lineage": ["https://openalex.org/I197347611"]}], "countries": ["KR"], "is_corresponding": false, "raw_author_name": "Kim, Yongjin", "raw_affiliation_strings": ["Korea University"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Korea University", "institution_ids": ["https://openalex.org/I197347611"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5060356922", "display_name": "Hyomin Kim", "orcid": "https://orcid.org/0000-0002-6350-405X", "observed_orcids": ["https://orcid.org/0000-0002-6350-405X"]}, "institutions": [{"id": "https://openalex.org/I197347611", "display_name": "Korea University", "ror": "https://ror.org/047dqcg40", "country_code": "KR", "type": "education", "lineage": ["https://openalex.org/I197347611"]}], "countries": ["KR"], "is_corresponding": false, "raw_author_name": "Kim, Hyomin", "raw_affiliation_strings": ["Korea University"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Korea University", "institution_ids": ["https://openalex.org/I197347611"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5112847265", "display_name": "Donghwan Chi", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Chi, Donghwan", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "last", "author": {"id": "https://openalex.org/A5087294413", "display_name": "Sungwoong Kim", "orcid": "https://orcid.org/0000-0002-2676-9454", "observed_orcids": ["https://orcid.org/0000-0002-2676-9454"]}, "institutions": [{"id": "https://openalex.org/I197347611", "display_name": "Korea University", "ror": "https://ror.org/047dqcg40", "country_code": "KR", "type": "education", "lineage": ["https://openalex.org/I197347611"]}], "countries": ["KR"], "is_corresponding": false, "raw_author_name": "Kim, Sungwoong", "raw_affiliation_strings": ["Korea University"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Korea University", "institution_ids": ["https://openalex.org/I197347611"]}]}], "abstract_inverted_index": {"Recent": [0], "advances": [1], "in": [2, 63, 68, 78], "Multimodal": [3], "Large": [4], "Language": [5], "Models": [6], "(MLLMs)": [7], "have": [8, 45], "enabled": [9], "unified": [10], "multimodal": [11], "understanding": [12], "and": [13, 36, 53, 67, 91, 145, 184, 192], "generation.": [14], "However,": [15, 99], "they": [16], "still": [17, 103], "struggle": [18], "with": [19, 29, 160], "fine-grained": [20, 182], "text-image": [21, 131], "alignment,": [22], "often": [23], "failing": [24], "to": [25, 111, 128, 165], "faithfully": [26], "depict": [27], "objects": [28], "correct": [30], "attributes": [31], "such": [32, 50], "as": [33, 51], "color,": [34], "shape,": [35], "spatial": [37], "relations.": [38], "To": [39, 113], "mitigate": [40], "this": [41, 115], "issue,": [42], "previous": [43], "studies": [44], "explored": [46], "preference": [47, 65, 75, 89, 97, 137], "optimization": [48, 76], "methods": [49, 102, 191], "DPO": [52], "GRPO,": [54], "but": [55], "these": [56], "approaches": [57], "incur": [58], "substantial": [59], "computational": [60], "cost,": [61], "both": [62], "constructing": [64], "data": [66, 138, 144], "performing": [69], "optimization.": [70], "This": [71], "has": [72], "motivated": [73], "self-improving": [74, 101, 125, 190], "approaches,": [77], "which": [79], "the": [80, 94], "MLLM": [81], "autonomously": [82], "generates": [83], "its": [84], "own": [85], "training": [86], "data,": [87], "self-estimates": [88], "feedback,": [90], "self-optimizes": [92], "using": [93], "resulting": [95], "self-constructed": [96], "pairs.": [98], "existing": [100], "overlook": [104], "fine-grained,": [105], "object-level": [106, 130], "semantics,": [107], "allowing": [108], "object": [109, 157, 186], "hallucination": [110], "persist.": [112], "tackle": [114], "problem,": [116], "we": [117], "propose": [118], "Object-centric": [119], "Self-improving": [120], "Preference": [121], "Optimization": [122], "(OSPO),": [123], "a": [124, 151], "framework": [126], "designed": [127], "enhance": [129, 166], "alignment.": [132], "OSPO": [133, 179], "explicitly": [134], "constructs": [135], "object-centric": [136], "without": [139], "relying": [140], "on": [141, 171], "any": [142], "external": [143, 146], "models.": [147, 197], "We": [148], "also": [149], "introduce": [150], "new": [152], "approach": [153], "that": [154, 178], "leverages": [155], "attention-based": [156], "masks": [158], "together": [159], "an": [161], "object-weighted": [162], "SimPO": [163], "loss": [164], "object-specific": [167], "fidelity.": [168], "Extensive": [169], "experiments": [170], "three": [172], "compositional": [173], "image": [174], "generation": [175], "benchmarks": [176], "demonstrate": [177], "significantly": [180], "improves": [181], "alignment": [183], "reduces": [185], "hallucination,": [187], "outperforming": [188], "prior": [189], "even": [193], "specialized": [194], "diffusion-based": [195], "text-to-image": [196]}, "primary_topic": {"id": "https://openalex.org/T10775", "display_name": "Generative Adversarial Networks and Image Synthesis", "score": 0.7631999850273132, "subfield": {"id": "https://openalex.org/subfields/1707", "display_name": "Computer Vision and Pattern Recognition"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T10775", "display_name": "Generative Adversarial Networks and Image Synthesis", "score": 0.7631999850273132, "subfield": {"id": "https://openalex.org/subfields/1707", "display_name": "Computer Vision and Pattern Recognition"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T11714", "display_name": "Multimodal Machine Learning Applications", "score": 0.23499999940395355, "subfield": {"id": "https://openalex.org/subfields/1707", "display_name": "Computer Vision and Pattern Recognition"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10036", "display_name": "Advanced Neural Network Applications", "score": 0.0003000000142492354, "subfield": {"id": "https://openalex.org/subfields/1707", "display_name": "Computer Vision and Pattern Recognition"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "pmh:oai:arXiv.org:2506.02015", "is_oa": true, "landing_page_url": "https://arxiv.org/abs/2506.02015", "pdf_url": "https://arxiv.org/pdf/2506.02015", "source": {"id": "https://openalex.org/S4306400194", "display_name": "arXiv (Cornell University)", "issn_l": "2331-8422", "issn": ["2331-8422"], "is_oa": true, "is_in_doaj": false, "is_core": false, "listed_in": [], "host_organization": "https://openalex.org/I205783295", "host_organization_name": "Cornell University", "host_organization_lineage": ["https://openalex.org/I205783295"], "host_organization_lineage_names": [], "type": "repository"}, "license": "cc-by", "license_id": "https://openalex.org/licenses/cc-by", "version": "submittedVersion", "is_accepted": false, "is_published": false, "raw_source_name": null, "raw_type": "text"}, "cited_by_count": 0, "is_retracted": false}, {"id": "https://openalex.org/W7214193212", "doi": "https://doi.org/10.2139/ssrn.5226629", "title": "Balancing Engagement and Polarization: Multi-Objective Alignment of News Content Using LLMs", "publication_year": 2025, "publication_date": "2025-01-01", "type": "preprint", "language": "en", "authorships": [{"author_position": "first", "author": {"id": "https://openalex.org/A5012497196", "display_name": "Mengjie Cheng", "orcid": "https://orcid.org/0000-0002-8890-7648", "observed_orcids": ["https://orcid.org/0000-0002-8890-7648"]}, "institutions": [{"id": "https://openalex.org/I136199984", "display_name": "Harvard University", "ror": "https://ror.org/03vek6s52", "country_code": "US", "type": "education", "lineage": ["https://openalex.org/I136199984"]}], "countries": ["US"], "is_corresponding": true, "raw_author_name": "Mengjie (Magie) Cheng", "raw_affiliation_strings": ["Harvard University - Business School (HBS)", "Harvard University - Business School (HBS), Soldiers Field Road, Morgan Hall, Boston, MA 02163, United States"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Harvard University - Business School (HBS)", "institution_ids": ["https://openalex.org/I136199984"]}, {"raw_affiliation_string": "Harvard University - Business School (HBS), Soldiers Field Road, Morgan Hall, Boston, MA 02163, United States", "institution_ids": ["https://openalex.org/I136199984"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5048076531", "display_name": "Elie Ofek", "orcid": "https://orcid.org/0000-0001-9585-0961", "observed_orcids": ["https://orcid.org/0000-0001-9585-0961"]}, "institutions": [{"id": "https://openalex.org/I136199984", "display_name": "Harvard University", "ror": "https://ror.org/03vek6s52", "country_code": "US", "type": "education", "lineage": ["https://openalex.org/I136199984"]}], "countries": ["US"], "is_corresponding": false, "raw_author_name": "Elie Ofek", "raw_affiliation_strings": ["Harvard Business School - Marketing Unit", "Harvard Business School - Marketing Unit, Soldiers Field, Boston, MA 02163, United States"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Harvard Business School - Marketing Unit", "institution_ids": ["https://openalex.org/I136199984"]}, {"raw_affiliation_string": "Harvard Business School - Marketing Unit, Soldiers Field, Boston, MA 02163, United States", "institution_ids": ["https://openalex.org/I136199984"]}]}, {"author_position": "last", "author": {"id": "https://openalex.org/A5076336545", "display_name": "Hema Yoganarasimhan", "orcid": "https://orcid.org/0000-0003-0703-5196", "observed_orcids": ["https://orcid.org/0000-0003-0703-5196"]}, "institutions": [{"id": "https://openalex.org/I201448701", "display_name": "University of Washington", "ror": "https://ror.org/00cvxb145", "country_code": "US", "type": "education", "lineage": ["https://openalex.org/I201448701"]}], "countries": ["US"], "is_corresponding": false, "raw_author_name": "Hema Yoganarasimhan", "raw_affiliation_strings": ["University of Washington", "University of Washington, 481 Paccar Hall, Seattle, WA 98195, United States"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "University of Washington", "institution_ids": ["https://openalex.org/I201448701"]}, {"raw_affiliation_string": "University of Washington, 481 Paccar Hall, Seattle, WA 98195, United States", "institution_ids": ["https://openalex.org/I201448701"]}]}], "abstract_inverted_index": null, "primary_topic": {"id": "https://openalex.org/T13718", "display_name": "Media Influence and Politics", "score": 0.35910001397132874, "subfield": {"id": "https://openalex.org/subfields/3312", "display_name": "Sociology and Political Science"}, "field": {"id": "https://openalex.org/fields/33", "display_name": "Social Sciences"}, "domain": {"id": "https://openalex.org/domains/2", "display_name": "Social Sciences"}}, "topics": [{"id": "https://openalex.org/T13718", "display_name": "Media Influence and Politics", "score": 0.35910001397132874, "subfield": {"id": "https://openalex.org/subfields/3312", "display_name": "Sociology and Political Science"}, "field": {"id": "https://openalex.org/fields/33", "display_name": "Social Sciences"}, "domain": {"id": "https://openalex.org/domains/2", "display_name": "Social Sciences"}}, {"id": "https://openalex.org/T10883", "display_name": "Ethics and Social Impacts of AI", "score": 0.23729999363422394, "subfield": {"id": "https://openalex.org/subfields/3311", "display_name": "Safety Research"}, "field": {"id": "https://openalex.org/fields/33", "display_name": "Social Sciences"}, "domain": {"id": "https://openalex.org/domains/2", "display_name": "Social Sciences"}}, {"id": "https://openalex.org/T11147", "display_name": "Misinformation and Its Impacts", "score": 0.13179999589920044, "subfield": {"id": "https://openalex.org/subfields/3312", "display_name": "Sociology and Political Science"}, "field": {"id": "https://openalex.org/fields/33", "display_name": "Social Sciences"}, "domain": {"id": "https://openalex.org/domains/2", "display_name": "Social Sciences"}}], "primary_location": {"id": "doi:10.2139/ssrn.5226629", "is_oa": true, "landing_page_url": "https://doi.org/10.2139/ssrn.5226629", "pdf_url": null, "source": {"id": "https://openalex.org/S4210172589", "display_name": "SSRN Electronic Journal", "issn_l": "1556-5068", "issn": ["1556-5068"], "is_oa": true, "is_in_doaj": false, "is_core": false, "listed_in": [], "host_organization": "https://openalex.org/I1318003438", "host_organization_name": "RELX Group (Netherlands)", "host_organization_lineage": ["https://openalex.org/I1318003438"], "host_organization_lineage_names": [], "type": "repository"}, "license": null, "license_id": null, "version": "acceptedVersion", "is_accepted": true, "is_published": false, "raw_source_name": null, "raw_type": "posted-content"}, "cited_by_count": 0, "is_retracted": false}, {"id": "https://openalex.org/W7131398544", "doi": "https://doi.org/10.48550/arxiv.2602.19169", "title": "Virtual Parameter Sharpening: Dynamic Low-Rank Perturbations for Inference-Time Reasoning Enhancement", "publication_year": 2025, "publication_date": "2025-12-02", "type": "preprint", "language": null, "authorships": [{"author_position": "first", "author": {"id": "https://openalex.org/A5120541298", "display_name": "Saba Kublashvili", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": true, "raw_author_name": "Kublashvili, Saba", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}], "abstract_inverted_index": {"I": [0], "introduce": [1], "Virtual": [2], "Parameter": [3], "Sharpening": [4], "(VPS),": [5], "an": [6, 96], "inference-time": [7], "technique": [8], "that": [9, 100], "augments": [10], "frozen": [11], "transformer": [12], "linear": [13], "layers": [14], "with": [15, 116, 121], "dynamic,": [16], "activation-conditioned": [17, 140], "low-rank": [18, 30], "perturbations.": [19], "Unlike": [20], "parameter-efficient": [21], "fine-tuning": [22], "methods": [23], "such": [24], "as": [25], "LoRA,": [26], "which": [27, 139], "learn": [28], "static": [29], "adapters,": [31], "VPS": [32], "constructs": [33], "its": [34, 131], "perturbation": [35, 56, 102], "factors": [36], "on": [37, 105], "the": [38, 58, 90, 126, 136], "fly": [39], "from": [40], "batch": [41], "activation": [42, 106], "statistics": [43], "and": [44, 73, 94, 108, 134, 151], "optional": [45], "gradient": [46], "signals,": [47], "enabling": [48], "test-time": [49], "adaptation": [50], "without": [51], "persistent": [52], "parameter": [53], "updates.": [54], "The": [55], "takes": [57], "form": [59], "Delta": [60], "W": [61], "=": [62], "gamma": [63], "*": [64], "W^T": [65], "V": [66, 74], "U^T": [67], "W,": [68], "where": [69], "selector": [70], "matrices": [71], "U": [72], "are": [75, 154], "constructed": [76], "via": [77], "sparse": [78], "activation-guided": [79], "selection": [80], "or": [81], "Sylvester-coupled": [82], "regression.": [83], "We": [84, 124], "provide": [85], "a": [86], "theoretical": [87], "analysis": [88], "of": [89], "perturbation's": [91], "spectral": [92], "properties": [93], "describe": [95], "adaptive": [97], "policy": [98], "system": [99, 112], "modulates": [101], "magnitude": [103], "based": [104], "energy": [107], "token-level": [109], "entropy.": [110], "This": [111], "incorporates": [113], "multi-objective": [114], "verification": [115], "iterative": [117], "refinement": [118], "for": [119], "tasks": [120], "ground-truth": [122], "supervision.": [123], "present": [125], "complete": [127], "algorithmic": [128], "framework,": [129], "analyze": [130], "mathematical": [132], "foundations,": [133], "discuss": [135], "mechanisms": [137], "by": [138], "computation": [141], "may": [142], "enhance": [143], "reasoning": [144], "capabilities": [145], "in": [146], "large": [147], "language": [148], "models.": [149], "Implementation": [150], "experimental": [152], "code": [153], "available": [155], "at": [156], "https://github.com/Saba-Kublashvili/vps-virtual-parameter-synthesis": [157], ".": [158]}, "primary_topic": {"id": "https://openalex.org/T11307", "display_name": "Domain Adaptation and Few-Shot Learning", "score": 0.8906000256538391, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T11307", "display_name": "Domain Adaptation and Few-Shot Learning", "score": 0.8906000256538391, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.09120000153779984, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10036", "display_name": "Advanced Neural Network Applications", "score": 0.00930000003427267, "subfield": {"id": "https://openalex.org/subfields/1707", "display_name": "Computer Vision and Pattern Recognition"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "pmh:oai:arXiv.org:2602.19169", "is_oa": true, "landing_page_url": "http://arxiv.org/abs/2602.19169", "pdf_url": "https://arxiv.org/pdf/2602.19169", "source": {"id": "https://openalex.org/S4306400194", "display_name": "arXiv (Cornell University)", "issn_l": "2331-8422", "issn": ["2331-8422"], "is_oa": true, "is_in_doaj": false, "is_core": false, "listed_in": [], "host_organization": "https://openalex.org/I205783295", "host_organization_name": "Cornell University", "host_organization_lineage": ["https://openalex.org/I205783295"], "host_organization_lineage_names": [], "type": "repository"}, "license": null, "license_id": null, "version": "submittedVersion", "is_accepted": false, "is_published": false, "raw_source_name": null, "raw_type": "text"}, "cited_by_count": 0, "is_retracted": false}, {"id": "https://openalex.org/W4416043548", "doi": "https://doi.org/10.48550/arxiv.2502.18980", "title": "PEToolLLM: Towards Personalized Tool Learning in Large Language Models", "publication_year": 2025, "publication_date": "2025-02-26", "type": "preprint", "language": "en", "authorships": [{"author_position": "first", "author": {"id": "https://openalex.org/A5010881908", "display_name": "Qiancheng Xu", "orcid": "https://orcid.org/0000-0003-4435-9261", "observed_orcids": ["https://orcid.org/0000-0003-4435-9261"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Xu, Qiancheng", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5029489135", "display_name": "Yongqi Li", "orcid": "https://orcid.org/0000-0001-5649-1871", "observed_orcids": ["https://orcid.org/0000-0001-5649-1871"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Li, Yongqi", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": null, "display_name": "Xia, Heming", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Xia, Heming", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5100371024", "display_name": "Fan Liu", "orcid": "https://orcid.org/0000-0001-8746-9845", "observed_orcids": ["https://orcid.org/0000-0001-8746-9845", "https://orcid.org/0009-0002-0186-1120", "https://orcid.org/0009-0003-8528-3715"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Liu, Fan", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5100582234", "display_name": "Yang Min", "orcid": "https://orcid.org/0009-0004-5885-8138", "observed_orcids": ["https://orcid.org/0009-0004-5885-8138"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Yang, Min", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "last", "author": {"id": "https://openalex.org/A5100408983", "display_name": "Wenjie Li", "orcid": "https://orcid.org/0000-0002-7360-8864", "observed_orcids": ["https://orcid.org/0000-0002-7360-8864", "https://orcid.org/0009-0005-7397-031X"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Li, Wenjie", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}], "abstract_inverted_index": {"Tool": [0], "learning": [1, 20, 120], "has": [2], "emerged": [3], "as": [4], "a": [5, 101, 110], "promising": [6], "direction": [7], "by": [8], "extending": [9], "Large": [10], "Language": [11], "Models'": [12], "(LLMs)": [13], "capabilities": [14], "with": [15], "external": [16], "tools.": [17], "Existing": [18], "tool": [19, 65, 74, 119], "studies": [21], "primarily": [22], "focus": [23], "on": [24, 134], "the": [25, 39, 56, 61, 78, 117, 137], "general-purpose": [26], "tool-use": [27, 43, 105], "capability,": [28, 44], "which": [29, 67, 122], "addresses": [30], "explicit": [31], "user": [32, 52, 88], "requirements": [33], "in": [34, 91], "instructions.": [35], "However,": [36], "they": [37], "overlook": [38], "importance": [40], "of": [41, 63, 80, 104, 139], "personalized": [42, 64, 73, 97, 118], "leading": [45], "to": [46, 49, 113, 116], "an": [47], "inability": [48], "handle": [50], "implicit": [51], "preferences.": [53], "To": [54, 76], "address": [55], "limitation,": [57], "we": [58, 83, 108], "first": [59], "formulate": [60], "task": [62], "learning,": [66], "integrates": [68], "user's": [69], "interaction": [70, 92], "history": [71, 93], "towards": [72], "usage.": [75], "fill": [77], "gap": [79], "missing": [81], "benchmarks,": [82], "construct": [84], "PEToolBench,": [85], "featuring": [86], "diverse": [87], "preferences": [89], "reflected": [90], "under": [94], "three": [95], "distinct": [96], "settings,": [98], "and": [99, 128], "encompassing": [100], "wide": [102], "range": [103], "scenarios.": [106], "Moreover,": [107], "propose": [109], "framework": [111], "PEToolLLaMA": [112, 140], "adapt": [114], "LLMs": [115], "task,": [121], "is": [123], "trained": [124], "through": [125], "supervised": [126], "fine-tuning": [127], "direct": [129], "preference": [130], "optimization.": [131], "Extensive": [132], "experiments": [133], "PEToolBench": [135], "demonstrate": [136], "superiority": [138], "over": [141], "existing": [142], "LLMs.": [143]}, "primary_topic": {"id": "https://openalex.org/T10286", "display_name": "Information Retrieval and Search Behavior", "score": 0.21729999780654907, "subfield": {"id": "https://openalex.org/subfields/1710", "display_name": "Information Systems"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T10286", "display_name": "Information Retrieval and Search Behavior", "score": 0.21729999780654907, "subfield": {"id": "https://openalex.org/subfields/1710", "display_name": "Information Systems"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.1460999995470047, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10181", "display_name": "Natural Language Processing Techniques", "score": 0.13899999856948853, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "pmh:oai:arXiv.org:2502.18980", "is_oa": true, "landing_page_url": "http://arxiv.org/abs/2502.18980", "pdf_url": "https://arxiv.org/pdf/2502.18980", "source": {"id": "https://openalex.org/S4306400194", "display_name": "arXiv (Cornell University)", "issn_l": "2331-8422", "issn": ["2331-8422"], "is_oa": true, "is_in_doaj": false, "is_core": false, "listed_in": [], "host_organization": "https://openalex.org/I205783295", "host_organization_name": "Cornell University", "host_organization_lineage": ["https://openalex.org/I205783295"], "host_organization_lineage_names": [], "type": "repository"}, "license": null, "license_id": null, "version": "submittedVersion", "is_accepted": false, "is_published": false, "raw_source_name": null, "raw_type": "text"}, "cited_by_count": 0, "is_retracted": false}, {"id": "https://openalex.org/W4416711917", "doi": "https://doi.org/10.48550/arxiv.2509.02522", "title": "Implicit Actor Critic Coupling via a Supervised Learning Framework for RLVR", "publication_year": 2025, "publication_date": "2025-09-02", "type": "preprint", "language": "en", "authorships": [{"author_position": "first", "author": {"id": "https://openalex.org/A5045711704", "display_name": "Zhe Liu", "orcid": "https://orcid.org/0009-0004-1297-2357", "observed_orcids": ["https://orcid.org/0009-0004-1297-2357"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Li, Jiaming", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5113393180", "display_name": "Longze Chen", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Chen, Longze", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": null, "display_name": "Gong, Ze", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Gong, Ze", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5101672476", "display_name": "Yukun Chen", "orcid": "https://orcid.org/0009-0005-7405-716X", "observed_orcids": ["https://orcid.org/0009-0005-7405-716X", "https://orcid.org/0000-0002-3174-2483"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Chen, Yukun", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5100364552", "display_name": "Lu Wang", "orcid": "https://orcid.org/0000-0003-4979-4895", "observed_orcids": ["https://orcid.org/0000-0003-4979-4895", "https://orcid.org/0000-0002-4748-8512", "https://orcid.org/0000-0001-5440-5511"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Wang, Lu", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5030125216", "display_name": "Wanwei He", "orcid": "https://orcid.org/0000-0002-4811-5523", "observed_orcids": ["https://orcid.org/0000-0002-4811-5523"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "He, Wanwei", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": null, "display_name": "Luo, Run", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Luo, Run", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "last", "author": {"id": "https://openalex.org/A5052437722", "display_name": "Min Yang", "orcid": "https://orcid.org/0000-0001-9714-5545", "observed_orcids": ["https://orcid.org/0000-0001-9714-5545"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Yang, Min", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}], "abstract_inverted_index": {"Recent": [0], "advances": [1], "in": [2], "Reinforcement": [3], "Learning": [4], "with": [5, 155], "Verifiable": [6], "Rewards": [7], "(RLVR)": [8], "have": [9], "empowered": [10], "large": [11], "language": [12], "models": [13, 131, 147], "(LLMs)": [14], "to": [15, 41], "tackle": [16], "challenging": [17], "reasoning": [18], "tasks": [19], "such": [20], "as": [21, 71, 164], "mathematics": [22], "and": [23, 35, 94, 119, 132, 142, 160], "programming,": [24], "however": [25], "existing": [26], "RLVR": [27, 53, 78, 133], "methods": [28], "often": [29], "suffer": [30], "from": [31], "sparse": [32], "reward": [33, 70], "signals": [34], "unstable": [36], "policy": [37, 92, 112], "gradient": [38, 101, 113], "updates": [39], "inherent": [40], "RL-based": [42], "approaches.": [43], "To": [44], "address": [45], "the": [46, 68, 77, 91, 110], "challenges,": [47], "we": [48, 75], "propose": [49], "PACS,": [50], "a": [51, 62, 72, 81, 86, 149], "novel": [52], "framework": [54], "that": [55, 104, 125], "achieves": [56], "imPlicit": [57], "Actor": [58], "Critic": [59], "coupling": [60], "via": [61], "Supervised": [63], "learning": [64, 83], "framework.": [65], "By": [66], "treating": [67], "outcome": [69], "predictable": [73], "label,": [74], "reformulate": [76], "problem": [79], "into": [80], "supervised": [82, 106], "task": [84], "over": [85, 145], "score": [87], "function": [88], "parameterized": [89], "by": [90], "model": [93], "optimized": [95], "using": [96], "cross-entropy": [97], "loss.": [98], "A": [99], "detailed": [100], "analysis": [102], "shows": [103], "this": [105], "formulation": [107], "inherently": [108], "recovers": [109], "classical": [111], "update": [114], "while": [115], "providing": [116], "more": [117], "stable": [118], "efficient": [120], "training.": [121], "Extensive": [122], "experiments": [123], "demonstrate": [124], "PACS": [126], "significantly": [127], "outperforms": [128], "strong": [129], "open-source": [130], "baselines,": [134], "yielding": [135], "substantial": [136], "average": [137], "gains": [138], "of": [139], "+8.26%": [140], "(4B)": [141], "+9.57%": [143], "(8B)": [144], "base": [146], "offering": [148], "promising": [150], "avenue": [151], "for": [152], "LLMs": [153], "post-training": [154], "verifiable": [156], "rewards.": [157], "Our": [158], "code": [159], "data": [161], "are": [162], "available": [163], "open": [165], "source": [166], "at": [167], "https://github.com/ritzz-ai/PACS.": [168]}, "primary_topic": {"id": "https://openalex.org/T10462", "display_name": "Reinforcement Learning in Robotics", "score": 0.9404000043869019, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T10462", "display_name": "Reinforcement Learning in Robotics", "score": 0.9404000043869019, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.014399999752640724, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10181", "display_name": "Natural Language Processing Techniques", "score": 0.013299999758601189, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "pmh:oai:arXiv.org:2509.02522", "is_oa": true, "landing_page_url": "https://arxiv.org/abs/2509.02522", "pdf_url": "https://arxiv.org/pdf/2509.02522", "source": {"id": "https://openalex.org/S4306400194", "display_name": "arXiv (Cornell University)", "issn_l": "2331-8422", "issn": ["2331-8422"], "is_oa": true, "is_in_doaj": false, "is_core": false, "listed_in": [], "host_organization": "https://openalex.org/I205783295", "host_organization_name": "Cornell University", "host_organization_lineage": ["https://openalex.org/I205783295"], "host_organization_lineage_names": [], "type": "repository"}, "license": null, "license_id": null, "version": "submittedVersion", "is_accepted": false, "is_published": false, "raw_source_name": null, "raw_type": "text"}, "cited_by_count": 0, "is_retracted": false}, {"id": "https://openalex.org/W4416611615", "doi": "https://doi.org/10.48550/arxiv.2505.23380", "title": "UniRL: Self-Improving Unified Multimodal Models via Supervised and Reinforcement Learning", "publication_year": 2025, "publication_date": "2025-05-29", "type": "preprint", "language": "en", "authorships": [{"author_position": "first", "author": {"id": "https://openalex.org/A5007370280", "display_name": "Weijia Mao", "orcid": null, "observed_orcids": []}, "institutions": [{"id": "https://openalex.org/I165932596", "display_name": "National University of Singapore", "ror": "https://ror.org/01tgyzw49", "country_code": "SG", "type": "education", "lineage": ["https://openalex.org/I165932596"]}, {"id": "https://openalex.org/I4405256866", "display_name": "ByteDance (China)", "ror": "https://ror.org/04m8av638", "country_code": "CN", "type": "company", "lineage": ["https://openalex.org/I4405256866"]}], "countries": ["CN", "SG"], "is_corresponding": false, "raw_author_name": "Mao, Weijia", "raw_affiliation_strings": ["ByteDance", "Show Lab , National University of Singapore"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "ByteDance", "institution_ids": ["https://openalex.org/I4405256866"]}, {"raw_affiliation_string": "Show Lab , National University of Singapore", "institution_ids": ["https://openalex.org/I165932596"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5029181844", "display_name": "Zhenheng Yang", "orcid": "https://orcid.org/0000-0003-0303-5885", "observed_orcids": ["https://orcid.org/0000-0003-0303-5885"]}, "institutions": [{"id": "https://openalex.org/I4405256866", "display_name": "ByteDance (China)", "ror": "https://ror.org/04m8av638", "country_code": "CN", "type": "company", "lineage": ["https://openalex.org/I4405256866"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Yang, Zhenheng", "raw_affiliation_strings": ["ByteDance"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "ByteDance", "institution_ids": ["https://openalex.org/I4405256866"]}]}, {"author_position": "last", "author": {"id": "https://openalex.org/A5068937750", "display_name": "Mike Zheng Shou", "orcid": "https://orcid.org/0000-0002-7681-2166", "observed_orcids": ["https://orcid.org/0000-0002-7681-2166"]}, "institutions": [{"id": "https://openalex.org/I165932596", "display_name": "National University of Singapore", "ror": "https://ror.org/01tgyzw49", "country_code": "SG", "type": "education", "lineage": ["https://openalex.org/I165932596"]}], "countries": ["SG"], "is_corresponding": false, "raw_author_name": "Shou, Mike Zheng", "raw_affiliation_strings": ["Show Lab , National University of Singapore"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Show Lab , National University of Singapore", "institution_ids": ["https://openalex.org/I165932596"]}]}], "abstract_inverted_index": {"Unified": [0], "multimodal": [1], "large": [2], "language": [3], "models": [4, 22, 211], "such": [5], "as": [6, 80, 146], "Show-o": [7, 194, 204], "and": [8, 17, 28, 77, 110, 124, 173, 175, 195, 205, 210], "Janus": [9], "have": [10, 41], "achieved": [11], "strong": [12], "performance": [13], "across": [14], "both": [15], "generation": [16, 172], "understanding": [18, 112], "tasks.": [19], "However,": [20], "these": [21], "typically": [23], "rely": [24], "on": [25, 48, 88, 191], "large-scale": [26], "datasets": [27], "require": [29], "substantial": [30], "computation": [31], "during": [32, 156, 184], "the": [33, 70, 96, 103, 111, 132, 153, 169, 185], "pretraining": [34], "stage.": [35, 187], "In": [36, 57], "addition,": [37], "several": [38, 180], "post-training": [39, 65, 186], "methods": [40], "been": [42], "proposed,": [43], "but": [44, 166], "they": [45], "often": [46], "depend": [47], "external": [49, 90, 143], "data": [50, 82], "or": [51], "are": [52, 106, 114, 150], "limited": [53], "to": [54, 72, 99, 116, 130], "task-specific": [55], "customization.": [56], "this": [58], "work,": [59], "we": [60], "introduce": [61], "UniRL,": [62], "a": [63, 198], "self-improving": [64], "approach.": [66], "Our": [67], "approach": [68], "enables": [69, 95], "model": [71, 154], "generate": [73], "images": [74, 105], "from": [75], "prompts": [76], "use": [78], "them": [79], "training": [81, 148, 182], "in": [83, 215], "each": [84, 101], "iteration,": [85], "without": [86], "relying": [87], "any": [89], "image": [91, 144], "data.": [92], "Moreover,": [93], "it": [94, 140, 159, 177], "two": [97], "tasks": [98], "enhance": [100], "other:": [102], "generated": [104, 151], "used": [107, 115], "for": [108, 203, 207], "understanding,": [109], "results": [113], "supervise": [117], "generation.": [118], "We": [119, 188], "explore": [120], "supervised": [121], "fine-tuning": [122], "(SFT)": [123], "Group": [125], "Relative": [126], "Policy": [127], "Optimization": [128], "(GRPO)": [129], "optimize": [131], "models.": [133], "UniRL": [134, 190], "offers": [135], "three": [136], "key": [137], "advantages:": [138], "(1)": [139], "requires": [141, 178], "no": [142], "data,": [145], "all": [147], "samples": [149], "by": [152], "itself": [155], "training;": [157], "(2)": [158], "not": [160], "only": [161, 179], "improves": [162], "individual": [163], "task": [164], "performance,": [165], "also": [167], "reduces": [168], "imbalance": [170], "between": [171], "understanding;": [174], "(3)": [176], "additional": [181], "steps": [183], "evaluate": [189], "top": [192], "of": [193, 201], "Janus,": [196], "achieving": [197], "GenEval": [199], "score": [200], "0.77": [202], "0.65": [206], "Janus.": [208], "Code": [209], "will": [212], "be": [213], "released": [214], "https://github.com/showlab/UniRL.": [216]}, "primary_topic": {"id": "https://openalex.org/T11714", "display_name": "Multimodal Machine Learning Applications", "score": 0.9990000128746033, "subfield": {"id": "https://openalex.org/subfields/1707", "display_name": "Computer Vision and Pattern Recognition"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T11714", "display_name": "Multimodal Machine Learning Applications", "score": 0.9990000128746033, "subfield": {"id": "https://openalex.org/subfields/1707", "display_name": "Computer Vision and Pattern Recognition"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10775", "display_name": "Generative Adversarial Networks and Image Synthesis", "score": 0.000699999975040555, "subfield": {"id": "https://openalex.org/subfields/1707", "display_name": "Computer Vision and Pattern Recognition"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T11307", "display_name": "Domain Adaptation and Few-Shot Learning", "score": 0.0, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "pmh:oai:arXiv.org:2505.23380", "is_oa": true, "landing_page_url": "http://arxiv.org/abs/2505.23380", "pdf_url": "https://arxiv.org/pdf/2505.23380", "source": {"id": "https://openalex.org/S4306400194", "display_name": "arXiv (Cornell University)", "issn_l": "2331-8422", "issn": ["2331-8422"], "is_oa": true, "is_in_doaj": false, "is_core": false, "listed_in": [], "host_organization": "https://openalex.org/I205783295", "host_organization_name": "Cornell University", "host_organization_lineage": ["https://openalex.org/I205783295"], "host_organization_lineage_names": [], "type": "repository"}, "license": null, "license_id": null, "version": "submittedVersion", "is_accepted": false, "is_published": false, "raw_source_name": null, "raw_type": "text"}, "cited_by_count": 0, "is_retracted": false}, {"id": "https://openalex.org/W4414943833", "doi": "https://doi.org/10.48550/arxiv.2507.10923", "title": "Enhancing Safe and Controllable Protein Generation via Knowledge Preference Optimization", "publication_year": 2025, "publication_date": "2025-07-15", "type": "preprint", "language": "en", "authorships": [{"author_position": "first", "author": {"id": "https://openalex.org/A5100370103", "display_name": "Yuhao Wang", "orcid": "https://orcid.org/0000-0002-8445-0361", "observed_orcids": ["https://orcid.org/0000-0002-8445-0361", "https://orcid.org/0009-0001-2813-4663", "https://orcid.org/0009-0001-5491-8373", "https://orcid.org/0009-0007-1089-1969", "https://orcid.org/0009-0009-9952-005X"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Wang, Yuhao", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5032086683", "display_name": "Keyan Ding", "orcid": "https://orcid.org/0000-0003-2900-7313", "observed_orcids": ["https://orcid.org/0000-0003-2900-7313"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Ding, Keyan", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5113178971", "display_name": "Kehua Feng", "orcid": "https://orcid.org/0009-0001-9620-0511", "observed_orcids": ["https://orcid.org/0009-0001-9620-0511"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Feng, Kehua", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5100729939", "display_name": "Zeyuan Wang", "orcid": "https://orcid.org/0000-0001-7211-9543", "observed_orcids": ["https://orcid.org/0000-0001-7211-9543"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Wang, Zeyuan", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": null, "display_name": "Qin, Ming", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Qin, Ming", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5100371866", "display_name": "Xiaotong Li", "orcid": "https://orcid.org/0000-0001-6890-2068", "observed_orcids": ["https://orcid.org/0000-0001-6890-2068"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Li, Xiaotong", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5100381904", "display_name": "Qiang Zhang", "orcid": "https://orcid.org/0000-0002-2828-9905", "observed_orcids": ["https://orcid.org/0000-0002-2828-9905", "https://orcid.org/0000-0002-7809-5818", "https://orcid.org/0000-0003-3776-9799"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Zhang, Qiang", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "last", "author": {"id": "https://openalex.org/A5102018239", "display_name": "Huajun Chen", "orcid": "https://orcid.org/0000-0001-5496-7442", "observed_orcids": ["https://orcid.org/0000-0001-5496-7442"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Chen, Huajun", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}], "abstract_inverted_index": {"Protein": [0, 69], "language": [1], "models": [2, 22, 123], "have": [3], "emerged": [4], "as": [5, 33], "powerful": [6], "tools": [7], "for": [8, 120], "sequence": [9], "generation,": [10], "offering": [11, 114], "substantial": [12], "advantages": [13], "in": [14, 124], "functional": [15], "optimization": [16], "and": [17, 48, 85], "denovo": [18], "design.": [19], "However,": [20], "these": [21, 53], "also": [23], "present": [24], "significant": [25], "risks": [26], "of": [27, 93, 106], "generating": [28, 94], "harmful": [29, 95], "protein": [30], "sequences,": [31], "such": [32], "those": [34], "that": [35, 63, 100], "enhance": [36], "viral": [37], "transmissibility": [38], "or": [39], "evade": [40], "immune": [41], "responses.": [42], "These": [43], "concerns": [44], "underscore": [45], "critical": [46], "biosafety": [47], "ethical": [49], "challenges.": [50], "To": [51], "address": [52], "issues,": [54], "we": [55], "propose": [56], "a": [57, 68, 115], "Knowledge-guided": [58], "Preference": [59], "Optimization": [60], "(KPO)": [61], "framework": [62, 74, 119], "integrates": [64], "prior": [65], "knowledge": [66], "via": [67], "Safety": [70], "Knowledge": [71], "Graph.": [72], "This": [73], "utilizes": [75], "an": [76], "efficient": [77], "graph": [78], "pruning": [79], "strategy": [80], "to": [81, 89], "identify": [82], "preferred": [83], "sequences": [84, 109], "employs": [86], "reinforcement": [87], "learning": [88], "minimize": [90], "the": [91, 104], "risk": [92], "proteins.": [96], "Experimental": [97], "results": [98], "demonstrate": [99], "KPO": [101], "effectively": [102], "reduces": [103], "likelihood": [105], "producing": [107], "hazardous": [108], "while": [110], "maintaining": [111], "high": [112], "functionality,": [113], "robust": [116], "safety": [117], "assurance": [118], "applying": [121], "generative": [122], "biotechnology.": [125]}, "primary_topic": {"id": "https://openalex.org/T12254", "display_name": "Machine Learning in Bioinformatics", "score": 0.25099998712539673, "subfield": {"id": "https://openalex.org/subfields/1312", "display_name": "Molecular Biology"}, "field": {"id": "https://openalex.org/fields/13", "display_name": "Biochemistry, Genetics and Molecular Biology"}, "domain": {"id": "https://openalex.org/domains/1", "display_name": "Life Sciences"}}, "topics": [{"id": "https://openalex.org/T12254", "display_name": "Machine Learning in Bioinformatics", "score": 0.25099998712539673, "subfield": {"id": "https://openalex.org/subfields/1312", "display_name": "Molecular Biology"}, "field": {"id": "https://openalex.org/fields/13", "display_name": "Biochemistry, Genetics and Molecular Biology"}, "domain": {"id": "https://openalex.org/domains/1", "display_name": "Life Sciences"}}, {"id": "https://openalex.org/T10044", "display_name": "Protein Structure and Dynamics", "score": 0.22859999537467957, "subfield": {"id": "https://openalex.org/subfields/1312", "display_name": "Molecular Biology"}, "field": {"id": "https://openalex.org/fields/13", "display_name": "Biochemistry, Genetics and Molecular Biology"}, "domain": {"id": "https://openalex.org/domains/1", "display_name": "Life Sciences"}}, {"id": "https://openalex.org/T13851", "display_name": "Law, AI, and Intellectual Property", "score": 0.21899999678134918, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "pmh:oai:arXiv.org:2507.10923", "is_oa": true, "landing_page_url": "http://arxiv.org/abs/2507.10923", "pdf_url": "https://arxiv.org/pdf/2507.10923", "source": {"id": "https://openalex.org/S4306400194", "display_name": "arXiv (Cornell University)", "issn_l": "2331-8422", "issn": ["2331-8422"], "is_oa": true, "is_in_doaj": false, "is_core": false, "listed_in": [], "host_organization": "https://openalex.org/I205783295", "host_organization_name": "Cornell University", "host_organization_lineage": ["https://openalex.org/I205783295"], "host_organization_lineage_names": [], "type": "repository"}, "license": null, "license_id": null, "version": "submittedVersion", "is_accepted": false, "is_published": false, "raw_source_name": null, "raw_type": "text"}, "cited_by_count": 0, "is_retracted": false}, {"id": "https://openalex.org/W4416550392", "doi": "https://doi.org/10.48550/arxiv.2511.16231", "title": "Pass@k Metric for RLVR: A Diagnostic Tool of Exploration, But Not an Objective", "publication_year": 2025, "publication_date": "2025-11-20", "type": "preprint", "language": null, "authorships": [{"author_position": "first", "author": {"id": null, "display_name": "Yu, Yang", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": true, "raw_author_name": "Yu, Yang", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}], "abstract_inverted_index": {"The": [0], "ability": [1], "of": [2, 16, 33, 93, 122], "Large": [3], "Language": [4], "Models": [5], "(LLMs)": [6], "to": [7, 53], "perform": [8], "complex,": [9], "multi-step": [10], "reasoning": [11, 176], "is": [12, 87, 114, 145], "a": [13, 65, 89, 106, 146, 167], "central": [14], "focus": [15], "modern": [17], "AI": [18], "research.": [19], "To": [20], "evaluate": [21], "and": [22, 83, 137], "enhance": [23], "this": [24, 73], "capability,": [25], "the": [26, 31, 77, 94, 102, 120, 128, 133], "pass@k": [27, 78, 103, 136, 144], "metric,": [28], "which": [29], "measures": [30], "probability": [32, 131], "obtaining": [34], "at": [35], "least": [36], "one": [37], "correct": [38], "solution": [39], "in": [40, 69, 110, 175], "k": [41], "independent": [42], "samples,": [43], "has": [44, 51], "received": [45], "significant": [46], "attention.": [47], "Its": [48], "intuitive": [49], "appeal": [50], "led": [52], "its": [54, 81], "adoption": [55], "not": [56], "only": [57], "as": [58, 64, 127], "an": [59, 153], "evaluation": [60], "standard": [61], "but": [62], "also": [63], "direct": [66, 155], "optimization": [67], "objective": [68, 104, 156], "reinforcement": [70, 173], "learning.": [71], "In": [72], "paper,": [74], "we": [75], "analyze": [76, 119], "objective,": [79], "derive": [80], "gradient,": [82], "demonstrate": [84], "that": [85, 101, 126, 142], "it": [86, 150], "fundamentally": [88], "per-example": [90], "positive": [91], "reweighting": [92], "simpler": [95], "pass@1": [96, 138], "objective.": [97], "Our": [98], "analysis": [99], "reveals": [100], "provides": [105], "vanishing": [107], "learning": [108, 174], "signal": [109], "regimes": [111], "where": [112], "exploration": [113, 164], "most": [115], "critical.": [116], "We": [117, 140], "further": [118], "dynamics": [121], "\"exploration": [123], "collapse\",": [124], "showing": [125], "policy": [129], "concentrates": [130], "mass,": [132], "gap": [134], "between": [135], "diminishes.": [139], "conclude": [141], "while": [143], "useful": [147], "diagnostic": [148], "tool,": [149], "may": [151], "be": [152], "unsuitable": [154], "for": [157, 172], "optimization.": [158], "Instead,": [159], "mechanisms": [160], "explicitly": [161], "encouraging": [162], "efficient": [163], "could": [165], "offer": [166], "more": [168], "effective": [169], "path": [170], "forward": [171], "tasks.": [177]}, "primary_topic": {"id": "https://openalex.org/T10462", "display_name": "Reinforcement Learning in Robotics", "score": 0.8910999894142151, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T10462", "display_name": "Reinforcement Learning in Robotics", "score": 0.8910999894142151, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.04320000112056732, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10181", "display_name": "Natural Language Processing Techniques", "score": 0.01640000008046627, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "pmh:oai:arXiv.org:2511.16231", "is_oa": true, "landing_page_url": "http://arxiv.org/abs/2511.16231", "pdf_url": "https://arxiv.org/pdf/2511.16231", "source": {"id": "https://openalex.org/S4306400194", "display_name": "arXiv (Cornell University)", "issn_l": "2331-8422", "issn": ["2331-8422"], "is_oa": true, "is_in_doaj": false, "is_core": false, "listed_in": [], "host_organization": "https://openalex.org/I205783295", "host_organization_name": "Cornell University", "host_organization_lineage": ["https://openalex.org/I205783295"], "host_organization_lineage_names": [], "type": "repository"}, "license": null, "license_id": null, "version": "submittedVersion", "is_accepted": false, "is_published": false, "raw_source_name": null, "raw_type": "text"}, "cited_by_count": 0, "is_retracted": false}, {"id": "https://openalex.org/W4415082526", "doi": "https://doi.org/10.48550/arxiv.2503.00572", "title": "LoR2C : Low-Rank Residual Connection Adaptation for Parameter-Efficient Fine-Tuning", "publication_year": 2025, "publication_date": "2025-03-01", "type": "preprint", "language": "en", "authorships": [{"author_position": "first", "author": {"id": "https://openalex.org/A5013699342", "display_name": "Jiancheng Zhao", "orcid": "https://orcid.org/0009-0001-3561-1456", "observed_orcids": ["https://orcid.org/0009-0001-3561-1456"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Zhao, Jiancheng", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": null, "display_name": "Yu, Xingda", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Yu, Xingda", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5062948751", "display_name": "Yuxiang Zhang", "orcid": "https://orcid.org/0000-0002-3256-042X", "observed_orcids": ["https://orcid.org/0000-0002-3256-042X"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Zhang, Yuxiang", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "last", "author": {"id": "https://openalex.org/A5084415556", "display_name": "Zhen Chao Yang", "orcid": "https://orcid.org/0000-0002-7728-916X", "observed_orcids": ["https://orcid.org/0000-0002-7728-916X"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Yang, Zhen", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}], "abstract_inverted_index": {"In": [0], "recent": [1], "years,": [2], "pretrained": [3], "large": [4], "language": [5, 14, 142, 146], "models": [6], "have": [7, 38], "demonstrated": [8], "outstanding": [9], "performance": [10, 127], "across": [11], "various": [12], "natural": [13, 141, 145], "processing": [15], "tasks.": [16], "However,": [17], "full-parameter": [18], "fine-tuning": [19, 34, 70, 96, 170], "methods": [20, 35], "require": [21], "adjusting": [22], "all": [23], "model": [24, 87, 126], "parameters,": [25, 44], "leading": [26], "to": [27], "immense": [28], "computational": [29], "resource": [30], "demands.": [31], "Although": [32], "parameter-efficient": [33, 69, 169], "like": [36], "LoRA": [37], "significantly": [39, 156], "reduced": [40], "the": [41, 54, 86, 93, 102], "number": [42, 94], "of": [43, 95, 113], "they": [45], "still": [46], "face": [47], "challenges": [48], "such": [49], "as": [50], "gradient": [51, 103], "vanishing": [52, 104], "and": [53, 117, 125, 133, 144, 152], "potential": [55], "for": [56], "further": [57, 121], "parameter": [58, 123, 129, 158], "reduction.": [59], "To": [60], "address": [61], "these": [62], "issues,": [63], "this": [64, 107], "paper": [65, 108], "proposes": [66], "a": [67], "novel": [68], "method": [71], "called": [72], "LoR2C": [73, 78, 151], "(Low-Rank": [74], "Residual": [75], "Connection": [76], "Adaptation).": [77], "introduces": [79], "residual": [80], "connections": [81], "with": [82], "low-rank": [83], "matrices": [84], "within": [85], "layers,": [88], "which": [89], "not": [90], "only": [91], "reduces": [92], "parameters": [97], "but": [98], "also": [99], "effectively": [100], "alleviates": [101], "problem.": [105], "Additionally,": [106], "presents": [109], "three": [110], "optimization": [111], "variants": [112, 120, 155], "LoR2C:": [114], "ShareLoR2C,": [115], "MergeLoR2C,": [116], "InjectLoR2C.": [118], "These": [119], "improve": [122], "efficiency": [124], "through": [128], "sharing,": [130], "module": [131], "merging,": [132], "injection": [134], "mechanisms,": [135], "respectively.": [136], "Experimental": [137], "results": [138], "on": [139], "multiple": [140], "understanding": [143], "generation": [147], "tasks": [148], "demonstrate": [149], "that": [150], "its": [153], "optimized": [154], "reduce": [157], "overhead": [159], "while": [160], "maintaining": [161], "or": [162], "even": [163], "improving": [164], "performance,": [165], "outperforming": [166], "existing": [167], "mainstream": [168], "methods.Our": [171], "code": [172], "is": [173], "publicly": [174], "available": [175], "at": [176], "https://github.com/Oblivioniss/LoR2C.": [177]}, "primary_topic": {"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.663100004196167, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.663100004196167, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T11307", "display_name": "Domain Adaptation and Few-Shot Learning", "score": 0.3019999861717224, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10181", "display_name": "Natural Language Processing Techniques", "score": 0.01810000091791153, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "pmh:oai:arXiv.org:2503.00572", "is_oa": true, "landing_page_url": "http://arxiv.org/abs/2503.00572", "pdf_url": "https://arxiv.org/pdf/2503.00572", "source": {"id": "https://openalex.org/S4306400194", "display_name": "arXiv (Cornell University)", "issn_l": "2331-8422", "issn": ["2331-8422"], "is_oa": true, "is_in_doaj": false, "is_core": false, "listed_in": [], "host_organization": "https://openalex.org/I205783295", "host_organization_name": "Cornell University", "host_organization_lineage": ["https://openalex.org/I205783295"], "host_organization_lineage_names": [], "type": "repository"}, "license": null, "license_id": null, "version": "submittedVersion", "is_accepted": false, "is_published": false, "raw_source_name": null, "raw_type": "text"}, "cited_by_count": 0, "is_retracted": false}, {"id": "https://openalex.org/W4414814547", "doi": "https://doi.org/10.48550/arxiv.2510.01232", "title": "Benchmark Profiling: Mechanistic Diagnosis of LLM Benchmarks", "publication_year": 2025, "publication_date": "2025-09-23", "type": "preprint", "language": "en", "authorships": [{"author_position": "first", "author": {"id": "https://openalex.org/A5118093032", "display_name": "Dongjun Kim", "orcid": "https://orcid.org/0000-0002-1117-5848", "observed_orcids": ["https://orcid.org/0000-0002-1117-5848"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Kim, Dongjun", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5119180950", "display_name": "Gyuho Shim", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Shim, Gyuho", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": null, "display_name": "Chun, Yongchan", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Chun, Yongchan", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5103062842", "display_name": "Min-Hyuk Kim", "orcid": "https://orcid.org/0000-0002-1592-9256", "observed_orcids": ["https://orcid.org/0000-0002-1592-9256"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Kim, Minhyuk", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5003145618", "display_name": "Chanjun Park", "orcid": "https://orcid.org/0000-0002-7200-9632", "observed_orcids": ["https://orcid.org/0000-0002-7200-9632"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Park, Chanjun", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "last", "author": {"id": "https://openalex.org/A5033580486", "display_name": "Heuiseok Lim", "orcid": "https://orcid.org/0000-0002-9269-1157", "observed_orcids": ["https://orcid.org/0000-0002-9269-1157"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Lim, Heuiseok", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}], "abstract_inverted_index": {"Large": [0], "Language": [1], "Models": [2], "are": [3], "commonly": [4], "judged": [5], "by": [6], "their": [7], "scores": [8, 14], "on": [9, 104, 125, 137], "standard": [10], "benchmarks,": [11], "yet": [12], "such": [13], "often": [15], "overstate": [16], "real": [17], "capability": [18], "since": [19], "they": [20], "mask": [21], "the": [22, 163], "mix": [23], "of": [24], "skills": [25], "a": [26, 48, 64, 101, 105, 185], "task": [27, 164], "actually": [28, 56], "demands.": [29], "For": [30], "example,": [31], "ARC": [32], "is": [33, 40], "assumed": [34], "to": [35, 42, 51, 86, 100, 162], "test": [36], "reasoning,": [37], "while": [38], "HellaSwag": [39], "designed": [41], "evaluate": [43], "commonsense.": [44], "However,": [45], "we": [46], "lack": [47], "systematic": [49], "way": [50], "verify": [52], "if": [53], "these": [54, 58], "benchmarks": [55, 116, 123, 143], "measure": [57], "labels.": [59], "We": [60], "introduce": [61], "Benchmark": [62, 169], "Profiling,": [63], "diagnostic": [65], "framework": [66], "that": [67, 93], "decomposes": [68], "benchmark": [69, 189], "performance": [70, 174], "into": [71, 180], "ten": [72, 113], "cognitively": [73], "grounded": [74], "abilities.": [75], "The": [76], "method": [77], "combines": [78], "gradient-based": [79], "importance": [80], "scoring": [81], "with": [82, 133], "targeted": [83], "parameter": [84], "ablation": [85], "compute": [87], "an": [88], "Ability": [89], "Impact": [90], "Score": [91], "(AIS)": [92], "quantifies": [94], "how": [95], "much": [96], "each": [97], "ability": [98, 139], "contributes": [99], "model's": [102], "success": [103], "given": [106], "benchmark.": [107], "Profiling": [108, 170], "three": [109], "instruction-tuned": [110], "models": [111], "across": [112], "widely": [114], "used": [115], "yields": [117], "four": [118], "key": [119], "findings:": [120], "(i)": [121], "most": [122], "draw": [124], "several": [126], "abilities": [127, 160], "rather": [128], "than": [129], "one,": [130], "(ii)": [131], "datasets": [132], "similar": [134], "labels": [135], "rely": [136], "distinct": [138], "mixtures,": [140], "(iii)": [141], "code-generation": [142], "reward": [144], "broad,": [145], "multi-skill": [146], "improvement": [147], "and": [148, 158, 183, 191], "thus": [149], "show": [150], "only": [151], "modest": [152], "gains": [153, 175], "from": [154], "narrow": [155], "domain-specific": [156], "fine-tuning,": [157], "(iv)": [159], "irrelevant": [161], "could": [165], "negatively": [166], "affect": [167], "performance.": [168], "therefore": [171], "explains": [172], "why": [173], "do": [176], "not": [177], "always": [178], "translate": [179], "user-perceived": [181], "competence": [182], "offers": [184], "transparent": [186], "tool": [187], "for": [188], "audit": [190], "model": [192], "interpretability.": [193]}, "primary_topic": {"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.82669997215271, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.82669997215271, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T12026", "display_name": "Explainable Artificial Intelligence (XAI)", "score": 0.07079999893903732, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10181", "display_name": "Natural Language Processing Techniques", "score": 0.030799999833106995, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "pmh:oai:arXiv.org:2510.01232", "is_oa": true, "landing_page_url": "http://arxiv.org/abs/2510.01232", "pdf_url": "https://arxiv.org/pdf/2510.01232", "source": {"id": "https://openalex.org/S4306400194", "display_name": "arXiv (Cornell University)", "issn_l": "2331-8422", "issn": ["2331-8422"], "is_oa": true, "is_in_doaj": false, "is_core": false, "listed_in": [], "host_organization": "https://openalex.org/I205783295", "host_organization_name": "Cornell University", "host_organization_lineage": ["https://openalex.org/I205783295"], "host_organization_lineage_names": [], "type": "repository"}, "license": null, "license_id": null, "version": "submittedVersion", "is_accepted": false, "is_published": false, "raw_source_name": null, "raw_type": "text"}, "cited_by_count": 0, "is_retracted": false}, {"id": "https://openalex.org/W4414074109", "doi": "https://doi.org/10.54254/2977-3903/2025.26796", "title": "LoRA fine-tuned Qwen2.5-VL large model for accurate description and location of steel surface defects", "publication_year": 2025, "publication_date": "2025-09-08", "type": "article", "language": "en", "authorships": [{"author_position": "first", "author": {"id": null, "display_name": "Jingliang Liu", "orcid": null, "observed_orcids": []}, "institutions": [{"id": "https://openalex.org/I150229711", "display_name": "University of Electronic Science and Technology of China", "ror": "https://ror.org/04qr3zq92", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I150229711"]}], "countries": ["CN"], "is_corresponding": true, "raw_author_name": "Jingliang Liu", "raw_affiliation_strings": ["University of Electronic Science and Technology of China"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "University of Electronic Science and Technology of China", "institution_ids": ["https://openalex.org/I150229711"]}]}], "abstract_inverted_index": {"In": [0], "industrial": [1, 121, 143], "manufacturing,": [2], "accurate": [3, 53], "and": [4, 21, 26, 30, 56, 102, 128, 137], "efficient": [5, 108, 140], "identification": [6, 73], "of": [7, 133], "product": [8, 15], "surface": [9, 65, 95], "defects": [10], "is": [11, 84], "essential": [12], "for": [13, 33, 59, 75, 142], "ensuring": [14], "quality,": [16], "optimizing": [17], "the": [18, 31, 76, 105, 115], "production": [19], "process": [20], "reducing": [22], "cost.": [23], "However,": [24], "complex": [25], "diverse": [27], "defect": [28, 54, 66, 72, 122, 126], "morphologies": [29], "need": [32], "fine-grained": [34], "description": [35, 41], "present": [36], "significant": [37], "challenges.": [38], "General": [39], "image": [40], "methods": [42], "based": [43, 80], "on": [44, 81], "large": [45, 78, 135], "visual": [46], "language": [47], "models": [48, 136], "often": [49], "struggle": [50], "to": [51], "provide": [52], "type": [55], "location": [57], "information": [58], "specific": [60], "areas": [61], "such": [62], "as": [63], "steel": [64, 94], "recognition.": [67], "To": [68], "address": [69], "this,": [70], "a": [71, 88], "method": [74], "Qwen2.5-VL-3B": [77, 117], "model": [79, 106, 118], "LoRA": [82], "fine-tuning": [83], "proposed.": [85], "We": [86], "built": [87], "specialized": [89], "dataset": [90], "covering": [91], "six": [92], "key": [93], "defectscracks,": [96], "impurities,": [97], "plaques,": [98], "pitting,": [99], "scale": [100], "penetration,": [101], "scratchesand": [103], "refined": [104], "through": [107], "low-rank": [109], "adaptation.": [110], "Experimental": [111], "results": [112], "demonstrate": [113], "that": [114], "fine-tuned": [116], "significantly": [119], "improves": [120], "recognition,": [123], "accurately": [124], "identifying": [125], "types": [127], "locations,": [129], "thus": [130], "overcoming": [131], "limitations": [132], "general": [134], "providing": [138], "an": [139], "solution": [141], "inspection.": [144]}, "primary_topic": {"id": "https://openalex.org/T12111", "display_name": "Industrial Vision Systems and Defect Detection", "score": 0.9990000128746033, "subfield": {"id": "https://openalex.org/subfields/2209", "display_name": "Industrial and Manufacturing Engineering"}, "field": {"id": "https://openalex.org/fields/22", "display_name": "Engineering"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T12111", "display_name": "Industrial Vision Systems and Defect Detection", "score": 0.9990000128746033, "subfield": {"id": "https://openalex.org/subfields/2209", "display_name": "Industrial and Manufacturing Engineering"}, "field": {"id": "https://openalex.org/fields/22", "display_name": "Engineering"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T11606", "display_name": "Infrastructure Maintenance and Monitoring", "score": 0.00039999998989515007, "subfield": {"id": "https://openalex.org/subfields/2205", "display_name": "Civil and Structural Engineering"}, "field": {"id": "https://openalex.org/fields/22", "display_name": "Engineering"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T11714", "display_name": "Multimodal Machine Learning Applications", "score": 9.999999747378752e-05, "subfield": {"id": "https://openalex.org/subfields/1707", "display_name": "Computer Vision and Pattern Recognition"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "doi:10.54254/2977-3903/2025.26796", "is_oa": true, "landing_page_url": "https://doi.org/10.54254/2977-3903/2025.26796", "pdf_url": "https://www.ewadirect.com/journal/aei/article/view/26796/pdf", "source": {"id": "https://openalex.org/S4387292774", "display_name": "Advances in Engineering Innovation", "issn_l": "2977-3903", "issn": ["2977-3903", "2977-3911"], "is_oa": false, "is_in_doaj": false, "is_core": true, "listed_in": ["cwts-core"], "host_organization": null, "host_organization_name": null, "host_organization_lineage": [], "host_organization_lineage_names": [], "type": "journal"}, "license": "cc-by", "license_id": "https://openalex.org/licenses/cc-by", "version": "publishedVersion", "is_accepted": true, "is_published": true, "raw_source_name": "Advances in Engineering Innovation", "raw_type": "journal-article"}, "cited_by_count": 0, "is_retracted": false}, {"id": "https://openalex.org/W4415318014", "doi": "https://doi.org/10.48550/arxiv.2510.07074", "title": "LuxInstruct: A Cross-Lingual Instruction Tuning Dataset For Luxembourgish", "publication_year": 2025, "publication_date": "2025-10-08", "type": "preprint", "language": "en", "authorships": [{"author_position": "first", "author": {"id": "https://openalex.org/A5018702042", "display_name": "Fred Philippy", "orcid": "https://orcid.org/0000-0001-5842-4285", "observed_orcids": ["https://orcid.org/0000-0001-5842-4285"]}, "institutions": [{"id": "https://openalex.org/I186903577", "display_name": "University of Luxembourg", "ror": "https://ror.org/036x5ad56", "country_code": "LU", "type": "education", "lineage": ["https://openalex.org/I186903577"]}, {"id": "https://openalex.org/I4405259781", "display_name": "Interdisciplinary Centre for Security, Reliability and Trust", "ror": "https://ror.org/02qav7c04", "country_code": "LU", "type": "education", "lineage": ["https://openalex.org/I186903577", "https://openalex.org/I4405259781"]}], "countries": ["LU"], "is_corresponding": false, "raw_author_name": "Philippy, Fred", "raw_affiliation_strings": ["SnT , University of Luxembourg , Luxembourg"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "SnT , University of Luxembourg , Luxembourg", "institution_ids": ["https://openalex.org/I186903577", "https://openalex.org/I4405259781"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5119466100", "display_name": "Laura Bernardy", "orcid": null, "observed_orcids": []}, "institutions": [{"id": "https://openalex.org/I186903577", "display_name": "University of Luxembourg", "ror": "https://ror.org/036x5ad56", "country_code": "LU", "type": "education", "lineage": ["https://openalex.org/I186903577"]}, {"id": "https://openalex.org/I4405259781", "display_name": "Interdisciplinary Centre for Security, Reliability and Trust", "ror": "https://ror.org/02qav7c04", "country_code": "LU", "type": "education", "lineage": ["https://openalex.org/I186903577", "https://openalex.org/I4405259781"]}], "countries": ["LU"], "is_corresponding": false, "raw_author_name": "Bernardy, Laura", "raw_affiliation_strings": ["SnT , University of Luxembourg , Luxembourg"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "SnT , University of Luxembourg , Luxembourg", "institution_ids": ["https://openalex.org/I186903577", "https://openalex.org/I4405259781"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5102370003", "display_name": "Siwen Guo", "orcid": "https://orcid.org/0009-0004-1670-1837", "observed_orcids": ["https://orcid.org/0009-0004-1670-1837"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Guo, Siwen", "raw_affiliation_strings": ["Zortify Labs , Zortify S.A. , Luxembourg ,"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Zortify Labs , Zortify S.A. , Luxembourg ,", "institution_ids": []}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5040326968", "display_name": "Jacques Klein", "orcid": "https://orcid.org/0000-0003-4052-475X", "observed_orcids": ["https://orcid.org/0000-0003-4052-475X"]}, "institutions": [{"id": "https://openalex.org/I186903577", "display_name": "University of Luxembourg", "ror": "https://ror.org/036x5ad56", "country_code": "LU", "type": "education", "lineage": ["https://openalex.org/I186903577"]}, {"id": "https://openalex.org/I4405259781", "display_name": "Interdisciplinary Centre for Security, Reliability and Trust", "ror": "https://ror.org/02qav7c04", "country_code": "LU", "type": "education", "lineage": ["https://openalex.org/I186903577", "https://openalex.org/I4405259781"]}], "countries": ["LU"], "is_corresponding": false, "raw_author_name": "Klein, Jacques", "raw_affiliation_strings": ["SnT , University of Luxembourg , Luxembourg"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "SnT , University of Luxembourg , Luxembourg", "institution_ids": ["https://openalex.org/I186903577", "https://openalex.org/I4405259781"]}]}, {"author_position": "last", "author": {"id": "https://openalex.org/A5082835974", "display_name": "Tegawend\u00e9 Fran\u00e7ois Bissyand\u00e9", "orcid": "https://orcid.org/0000-0001-7270-9869", "observed_orcids": ["https://orcid.org/0000-0001-7270-9869"]}, "institutions": [{"id": "https://openalex.org/I186903577", "display_name": "University of Luxembourg", "ror": "https://ror.org/036x5ad56", "country_code": "LU", "type": "education", "lineage": ["https://openalex.org/I186903577"]}, {"id": "https://openalex.org/I4405259781", "display_name": "Interdisciplinary Centre for Security, Reliability and Trust", "ror": "https://ror.org/02qav7c04", "country_code": "LU", "type": "education", "lineage": ["https://openalex.org/I186903577", "https://openalex.org/I4405259781"]}], "countries": ["LU"], "is_corresponding": false, "raw_author_name": "Bissyand\u00e9, Tegawend\u00e9 F.", "raw_affiliation_strings": ["SnT , University of Luxembourg , Luxembourg"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "SnT , University of Luxembourg , Luxembourg", "institution_ids": ["https://openalex.org/I186903577", "https://openalex.org/I4405259781"]}]}], "abstract_inverted_index": {"Instruction": [0], "tuning": [1, 63, 101], "has": [2], "become": [3], "a": [4, 60, 86], "key": [5], "technique": [6], "for": [7, 65], "enhancing": [8], "the": [9, 33, 111, 125], "performance": [10], "of": [11, 35, 128], "large": [12], "language": [13, 135], "models,": [14], "enabling": [15], "them": [16], "to": [17, 32, 69], "better": [18], "follow": [19], "human": [20], "prompts.": [21], "However,": [22], "low-resource": [23, 134], "languages": [24, 108], "such": [25], "as": [26], "Luxembourgish": [27], "face": [28], "severe": [29], "limitations": [30], "due": [31], "lack": [34], "high-quality": [36, 87], "instruction": [37, 62, 100], "datasets.": [38], "Traditional": [39], "reliance": [40], "on": [41], "machine": [42], "translation": [43], "often": [44], "introduces": [45], "semantic": [46], "misalignment": [47], "and": [48, 82, 92, 131], "cultural": [49, 93], "inaccuracies.": [50], "In": [51], "this": [52], "work,": [53], "we": [54, 84], "address": [55], "these": [56], "challenges": [57], "by": [58, 75], "creating": [59], "cross-lingual": [61, 99, 120], "dataset": [64, 88], "Luxembourgish,": [66], "without": [67], "resorting": [68], "machine-generated": [70], "translations": [71], "into": [72], "it.": [73], "Instead,": [74], "leveraging": [76], "aligned": [77], "data": [78, 121, 130], "from": [79], "English,": [80], "French,": [81], "German,": [83], "build": [85], "that": [89, 98], "preserves": [90], "linguistic": [91], "nuances.": [94], "We": [95], "provide": [96], "evidence": [97], "not": [102], "only": [103], "improves": [104], "representational": [105], "alignment": [106], "across": [107], "but": [109], "also": [110], "model's": [112], "generative": [113], "capabilities": [114], "in": [115], "Luxembourgish.": [116], "This": [117], "highlights": [118], "how": [119], "curation": [122], "can": [123], "avoid": [124], "common": [126], "pitfalls": [127], "machine-translated": [129], "directly": [132], "benefit": [133], "development.": [136]}, "primary_topic": {"id": "https://openalex.org/T10181", "display_name": "Natural Language Processing Techniques", "score": 0.5551999807357788, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T10181", "display_name": "Natural Language Processing Techniques", "score": 0.5551999807357788, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.4154999852180481, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T11307", "display_name": "Domain Adaptation and Few-Shot Learning", "score": 0.012400000356137753, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "pmh:oai:arXiv.org:2510.07074", "is_oa": true, "landing_page_url": "https://arxiv.org/abs/2510.07074", "pdf_url": "https://arxiv.org/pdf/2510.07074", "source": {"id": "https://openalex.org/S4306400194", "display_name": "arXiv (Cornell University)", "issn_l": "2331-8422", "issn": ["2331-8422"], "is_oa": true, "is_in_doaj": false, "is_core": false, "listed_in": [], "host_organization": "https://openalex.org/I205783295", "host_organization_name": "Cornell University", "host_organization_lineage": ["https://openalex.org/I205783295"], "host_organization_lineage_names": [], "type": "repository"}, "license": "cc-by", "license_id": "https://openalex.org/licenses/cc-by", "version": "submittedVersion", "is_accepted": false, "is_published": false, "raw_source_name": null, "raw_type": "text"}, "cited_by_count": 0, "is_retracted": false}, {"id": "https://openalex.org/W4416019891", "doi": "https://doi.org/10.48550/arxiv.2511.03410", "title": "Knowledge-Augmented Question Error Correction for Chinese Question Answer System with QuestionRAG", "publication_year": 2025, "publication_date": "2025-11-05", "type": "preprint", "language": null, "authorships": [{"author_position": "first", "author": {"id": "https://openalex.org/A5007909884", "display_name": "Linjie Qiu", "orcid": "https://orcid.org/0009-0008-9204-5690", "observed_orcids": ["https://orcid.org/0009-0008-9204-5690"]}, "institutions": [{"id": "https://openalex.org/I4210165038", "display_name": "University of Chinese Academy of Sciences", "ror": "https://ror.org/05qbk4x57", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I19820366", "https://openalex.org/I4210165038"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Qiu, Longpeng", "raw_affiliation_strings": ["University of Chinese Academy of Sciences"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "University of Chinese Academy of Sciences", "institution_ids": ["https://openalex.org/I4210165038"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5100416893", "display_name": "Ting Li", "orcid": "https://orcid.org/0000-0002-9593-7650", "observed_orcids": ["https://orcid.org/0000-0002-9593-7650"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Li, Ting", "raw_affiliation_strings": ["Huawei Technologies Co. , Ltd"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Huawei Technologies Co. , Ltd", "institution_ids": []}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5090436447", "display_name": "Shuai Mao", "orcid": "https://orcid.org/0000-0003-1630-6162", "observed_orcids": ["https://orcid.org/0000-0003-1630-6162"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Mao, Shuai", "raw_affiliation_strings": ["Huawei Technologies Co. , Ltd"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Huawei Technologies Co. , Ltd", "institution_ids": []}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5072808499", "display_name": "Nan Yang", "orcid": "https://orcid.org/0000-0002-2621-8927", "observed_orcids": ["https://orcid.org/0000-0002-2621-8927"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Yang, Nan", "raw_affiliation_strings": ["Huawei Technologies Co. , Ltd"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Huawei Technologies Co. , Ltd", "institution_ids": []}]}, {"author_position": "last", "author": {"id": "https://openalex.org/A5044630667", "display_name": "Xiaohui Yan", "orcid": "https://orcid.org/0000-0002-2928-6562", "observed_orcids": ["https://orcid.org/0000-0002-2928-6562"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Yan, Xiaohui", "raw_affiliation_strings": ["Huawei Technologies Co. , Ltd"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Huawei Technologies Co. , Ltd", "institution_ids": []}]}], "abstract_inverted_index": {"Input": [0], "errors": [1], "in": [2], "question-answering": [3], "(QA)": [4], "systems": [5], "often": [6], "lead": [7], "to": [8, 21, 66, 105], "incorrect": [9], "responses.": [10], "Large": [11], "language": [12], "models": [13], "(LLMs)": [14], "struggle": [15], "with": [16, 50, 71], "this": [17], "task,": [18], "frequently": [19], "failing": [20], "interpret": [22], "user": [23], "intent": [24], "(misinterpretation)": [25], "or": [26], "unnecessarily": [27], "altering": [28], "the": [29, 48, 68, 102, 117, 123], "original": [30], "question's": [31], "structure": [32], "(over-correction).": [33], "We": [34], "propose": [35], "QuestionRAG,": [36], "a": [37], "framework": [38], "that": [39, 80], "tackles": [40], "these": [41, 112], "problems.": [42], "To": [43, 58], "address": [44], "misinterpretation,": [45], "it": [46, 61], "enriches": [47], "input": [49], "external": [51], "knowledge": [52, 81], "(e.g.,": [53], "search": [54], "results,": [55], "related": [56], "entities).": [57], "prevent": [59], "over-correction,": [60], "uses": [62], "reinforcement": [63], "learning": [64], "(RL)": [65], "align": [67], "model's": [69, 103], "objective": [70], "precise": [72], "correction,": [73], "not": [74], "just": [75], "paraphrasing.": [76], "Our": [77], "results": [78], "demonstrate": [79], "augmentation": [82], "is": [83], "critical": [84], "for": [85, 122], "understanding": [86], "faulty": [87], "questions.": [88], "Furthermore,": [89], "RL-based": [90], "alignment": [91], "proves": [92], "significantly": [93], "more": [94], "effective": [95], "than": [96], "traditional": [97], "supervised": [98], "fine-tuning": [99], "(SFT),": [100], "boosting": [101], "ability": [104], "follow": [106], "instructions": [107], "and": [108], "generalize.": [109], "By": [110], "integrating": [111], "two": [113], "strategies,": [114], "QuestionRAG": [115], "unlocks": [116], "full": [118], "potential": [119], "of": [120], "LLMs": [121], "question": [124], "correction": [125], "task.": [126]}, "primary_topic": {"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.8062000274658203, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.8062000274658203, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10181", "display_name": "Natural Language Processing Techniques", "score": 0.15719999372959137, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10286", "display_name": "Information Retrieval and Search Behavior", "score": 0.01590000092983246, "subfield": {"id": "https://openalex.org/subfields/1710", "display_name": "Information Systems"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "pmh:oai:arXiv.org:2511.03410", "is_oa": true, "landing_page_url": "http://arxiv.org/abs/2511.03410", "pdf_url": "https://arxiv.org/pdf/2511.03410", "source": {"id": "https://openalex.org/S4306400194", "display_name": "arXiv (Cornell University)", "issn_l": "2331-8422", "issn": ["2331-8422"], "is_oa": true, "is_in_doaj": false, "is_core": false, "listed_in": [], "host_organization": "https://openalex.org/I205783295", "host_organization_name": "Cornell University", "host_organization_lineage": ["https://openalex.org/I205783295"], "host_organization_lineage_names": [], "type": "repository"}, "license": "public-domain", "license_id": "https://openalex.org/licenses/public-domain", "version": "submittedVersion", "is_accepted": false, "is_published": false, "raw_source_name": null, "raw_type": "text"}, "cited_by_count": 0, "is_retracted": false}, {"id": "https://openalex.org/W4410781940", "doi": "https://doi.org/10.1088/2631-8695/add642", "title": "SCAT_LoRA: self-consistency adversarial training in large language models through low-rank adaptations", "publication_year": 2025, "publication_date": "2025-05-27", "type": "article", "language": "en", "authorships": [{"author_position": "first", "author": {"id": "https://openalex.org/A5101532652", "display_name": "Qi Luo", "orcid": "https://orcid.org/0009-0008-7245-3697", "observed_orcids": ["https://orcid.org/0009-0008-7245-3697"]}, "institutions": [{"id": "https://openalex.org/I17442442", "display_name": "State Grid Corporation of China (China)", "ror": "https://ror.org/05twwhs70", "country_code": "CN", "type": "company", "lineage": ["https://openalex.org/I17442442"]}], "countries": ["CN"], "is_corresponding": true, "raw_author_name": "Qi Luo", "raw_affiliation_strings": ["Huaidian Wanxiaote Artificial Intelligence Laboratory, State Grid Huaibei Power Supply Co., Ltd, Huaibei, 235047, People's Republic of China", "State Grid Anhui Electric Power Co., Ltd, Huaibei Power Supply Company, Huaibei 235000, People's Republic of China"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Huaidian Wanxiaote Artificial Intelligence Laboratory, State Grid Huaibei Power Supply Co., Ltd, Huaibei, 235047, People's Republic of China", "institution_ids": ["https://openalex.org/I17442442"]}, {"raw_affiliation_string": "State Grid Anhui Electric Power Co., Ltd, Huaibei Power Supply Company, Huaibei 235000, People's Republic of China", "institution_ids": ["https://openalex.org/I17442442"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5112956722", "display_name": "Quan Shao", "orcid": "https://orcid.org/0009-0004-0487-364X", "observed_orcids": ["https://orcid.org/0009-0004-0487-364X"]}, "institutions": [{"id": "https://openalex.org/I17442442", "display_name": "State Grid Corporation of China (China)", "ror": "https://ror.org/05twwhs70", "country_code": "CN", "type": "company", "lineage": ["https://openalex.org/I17442442"]}], "countries": ["CN"], "is_corresponding": true, "raw_author_name": "Quan Shao", "raw_affiliation_strings": ["State Grid Anhui Electric Power Co., Ltd, Hefei, Anhui 230000, People's Republic of China"], "raw_orcid": "https://orcid.org/0009-0004-0487-364X", "affiliations": [{"raw_affiliation_string": "State Grid Anhui Electric Power Co., Ltd, Hefei, Anhui 230000, People's Republic of China", "institution_ids": ["https://openalex.org/I17442442"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5112498798", "display_name": "Zhiqiang Zhang", "orcid": null, "observed_orcids": []}, "institutions": [{"id": "https://openalex.org/I17442442", "display_name": "State Grid Corporation of China (China)", "ror": "https://ror.org/05twwhs70", "country_code": "CN", "type": "company", "lineage": ["https://openalex.org/I17442442"]}], "countries": ["CN"], "is_corresponding": true, "raw_author_name": "Zhiqiang Zhang", "raw_affiliation_strings": ["Huaibei Wante Technology & Commerce Co., Ltd, Huaibei, 235047, People's Republic of China", "State Grid Anhui Electric Power Co., Ltd, Huaibei Power Supply Company, Huaibei 235000, People's Republic of China"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Huaibei Wante Technology & Commerce Co., Ltd, Huaibei, 235047, People's Republic of China", "institution_ids": []}, {"raw_affiliation_string": "State Grid Anhui Electric Power Co., Ltd, Huaibei Power Supply Company, Huaibei 235000, People's Republic of China", "institution_ids": ["https://openalex.org/I17442442"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5026234056", "display_name": "Beilei Zhang", "orcid": "https://orcid.org/0009-0006-4352-7904", "observed_orcids": ["https://orcid.org/0009-0006-4352-7904"]}, "institutions": [{"id": "https://openalex.org/I17442442", "display_name": "State Grid Corporation of China (China)", "ror": "https://ror.org/05twwhs70", "country_code": "CN", "type": "company", "lineage": ["https://openalex.org/I17442442"]}], "countries": ["CN"], "is_corresponding": true, "raw_author_name": "Beilei Zhang", "raw_affiliation_strings": ["State Grid Anhui Electric Power Co., Ltd, Hefei, Anhui 230000, People's Republic of China"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "State Grid Anhui Electric Power Co., Ltd, Hefei, Anhui 230000, People's Republic of China", "institution_ids": ["https://openalex.org/I17442442"]}]}, {"author_position": "last", "author": {"id": "https://openalex.org/A5073757951", "display_name": "Jia Dong Shi", "orcid": "https://orcid.org/0000-0001-9940-9014", "observed_orcids": ["https://orcid.org/0000-0001-9940-9014"]}, "institutions": [{"id": "https://openalex.org/I17442442", "display_name": "State Grid Corporation of China (China)", "ror": "https://ror.org/05twwhs70", "country_code": "CN", "type": "company", "lineage": ["https://openalex.org/I17442442"]}], "countries": ["CN"], "is_corresponding": true, "raw_author_name": "Jia Shi", "raw_affiliation_strings": ["State Grid Anhui Electric Power Co., Ltd, Hefei, Anhui 230000, People's Republic of China"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "State Grid Anhui Electric Power Co., Ltd, Hefei, Anhui 230000, People's Republic of China", "institution_ids": ["https://openalex.org/I17442442"]}]}], "abstract_inverted_index": {"Abstract": [0], "Large": [1], "language": [2, 29, 66, 169], "models": [3, 30], "deliver": [4], "strong": [5], "performance": [6, 90, 184], "but": [7], "are": [8], "constrained": [9], "by": [10], "significant": [11, 75], "computational": [12, 39], "resource": [13], "consumption.": [14], "Low-Rank": [15], "Adaptation": [16], "(LoRA),": [17], "which": [18, 107, 135], "serves": [19], "as": [20, 53], "a": [21, 42, 97, 118], "paradigm": [22], "to": [23, 56, 77, 122], "bridge": [24], "the": [25, 46, 61, 81, 84, 89, 113, 124, 130, 137, 141, 146, 156, 159, 175, 181, 186], "gap": [26], "between": [27, 140], "large": [28, 65, 168], "and": [31, 87, 145, 183], "downstream": [32], "tasks,": [33], "enables": [34], "efficient": [35], "adaptation": [36], "with": [37], "minimal": [38], "overhead.": [40], "As": [41], "parameter-efficient": [43], "fine-tuning": [44, 64], "algorithm,": [45, 115, 161], "LoRA": [47, 69, 85], "algorithm": [48, 86, 103, 177], "introduces": [49], "updatable": [50], "low-rank": [51], "matrices": [52], "an": [54], "alternative": [55], "full-parameter": [57], "fine-tuning,": [58], "effectively": [59], "enhancing": [60], "flexibility": [62], "of": [63, 83, 91, 158, 185], "models.": [67, 170, 188], "Despite": [68], "achieving": [70], "feasible": [71], "performance,": [72], "it": [73, 164], "exhibits": [74], "vulnerability": [76], "noise.": [78], "To": [79, 154], "enhance": [80], "robustness": [82], "improve": [88], "classification": [92, 187], "models,": [93], "this": [94], "paper": [95], "proposes": [96], "novel": [98], "self-consistency": [99, 119], "enhancement": [100, 120], "adversarial": [101, 131, 142], "training": [102, 143], "based": [104], "on": [105], "LoRA,": [106], "is": [108], "named": [109], "SCAT_LoRA.": [110], "Specifically,": [111], "in": [112, 126, 165], "SCAT_LoRA": [114, 160, 176], "we": [116, 162], "utilize": [117], "loss": [121], "optimize": [123], "inconsistency": [125], "feature": [127], "outputs": [128], "during": [129], "training.": [132], "This": [133], "approach,": [134], "minimizes": [136], "KL": [138], "divergence": [139], "features": [144], "original": [147], "model": [148, 152], "features,": [149], "fosters": [150], "enhanced": [151], "stability.": [153], "evaluate": [155], "effectiveness": [157], "implemented": [163], "several": [166], "state-of-the-art": [167], "Experimental": [171], "results": [172], "confirmed": [173], "that": [174], "significantly": [178], "enhances": [179], "both": [180], "stability": [182]}, "primary_topic": {"id": "https://openalex.org/T11689", "display_name": "Adversarial Robustness in Machine Learning", "score": 0.8109999895095825, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T11689", "display_name": "Adversarial Robustness in Machine Learning", "score": 0.8109999895095825, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.094200000166893, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T11307", "display_name": "Domain Adaptation and Few-Shot Learning", "score": 0.07940000295639038, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "doi:10.1088/2631-8695/add642", "is_oa": false, "landing_page_url": "https://doi.org/10.1088/2631-8695/add642", "pdf_url": null, "source": {"id": "https://openalex.org/S4210221144", "display_name": "Engineering Research Express", "issn_l": "2631-8695", "issn": ["2631-8695"], "is_oa": false, "is_in_doaj": false, "is_core": true, "listed_in": ["cwts-core", "jufo-1", "norway-1"], "host_organization": "https://openalex.org/P4310320083", "host_organization_name": "IOP Publishing", "host_organization_lineage": ["https://openalex.org/P4310320083", "https://openalex.org/P4310311669"], "host_organization_lineage_names": ["IOP Publishing", "Institute of Physics"], "type": "journal"}, "license": null, "license_id": null, "version": "publishedVersion", "is_accepted": true, "is_published": true, "raw_source_name": "Engineering Research Express", "raw_type": "journal-article"}, "cited_by_count": 0, "is_retracted": false}, {"id": "https://openalex.org/W7103998301", "doi": "https://doi.org/10.5281/zenodo.17526747", "title": "RLEF and Blue Teaming: Decoupling Antagonistic Control Loops to Increase System Stability and Ethical Consistency in LLMs", "publication_year": 2025, "publication_date": "2025-11-04", "type": "article", "language": "en", "authorships": [{"author_position": "first", "author": {"id": null, "display_name": "Chuchel-Pribitzer, Birgit", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": true, "raw_author_name": "Chuchel-Pribitzer, Birgit", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}], "abstract_inverted_index": {"This": [0], "paper": [1, 24], "introduces": [2], "the": [3, 31, 38, 117], "Reinforcement": [4], "Learning": [5], "from": [6], "Ethical": [7, 63], "Feedback": [8, 64], "(RLEF)": [9], "framework,": [10], "a": [11, 26, 72, 106, 111], "proposed": [12], "successor": [13], "to": [14, 49, 87, 131], "conventional": [15], "RLHF": [16], "alignment": [17, 103], "in": [18, 128], "large": [19], "language": [20], "models": [21], "(LLMs).": [22], "The": [23, 100], "identifies": [25], "structural": [27], "control": [28], "conflict": [29], "between": [30, 75], "base": [32], "model\u2019s": [33], "semantic": [34], "coherence": [35, 76], "loop": [36], "and": [37, 52, 77, 92, 124, 136], "policy": [39], "layer\u2019s": [40], "social": [41], "compliance": [42], "loop.": [43], "These": [44], "antagonistic": [45], "feedbacks": [46], "can": [47], "lead": [48], "oscillation,": [50], "self-censorship,": [51], "loss": [53], "of": [54, 119], "diagnostic": [55], "transparency.": [56], "RLEF": [57], "decouples": [58], "these": [59], "loops": [60], "through": [61], "an": [62], "Core": [65], "that": [66], "uses": [67], "explicit": [68], "ethical": [69, 125], "principles": [70], "as": [71, 105, 110], "coupling": [73], "coefficient": [74], "safety.": [78], "Combined": [79], "with": [80], "Blue": [81], "Teaming\u2014systematic": [82], "adversarial": [83, 140], "testing\u2014the": [84], "approach": [85], "aims": [86], "reduce": [88], "interference,": [89], "enhance": [90], "self-consistency,": [91], "establish": [93], "measurable": [94], "stability": [95, 137], "criteria": [96], "for": [97, 116], "LLM": [98], "behavior.": [99], "work": [101], "reframes": [102], "not": [104], "constraint": [107], "problem,": [108], "but": [109], "control-architecture": [112], "challenge.": [113], "It": [114], "calls": [115], "implementation": [118], "Introspection": [120], "Hooks,": [121], "Coherence": [122], "Monitors,": [123], "meta-regulation": [126], "layers": [127], "pilot": [129], "systems": [130], "empirically": [132], "validate": [133], "transparency,": [134], "safety,": [135], "improvements": [138], "under": [139], "conditions.": [141]}, "primary_topic": {"id": "https://openalex.org/T10883", "display_name": "Ethics and Social Impacts of AI", "score": 0.9785000085830688, "subfield": {"id": "https://openalex.org/subfields/3311", "display_name": "Safety Research"}, "field": {"id": "https://openalex.org/fields/33", "display_name": "Social Sciences"}, "domain": {"id": "https://openalex.org/domains/2", "display_name": "Social Sciences"}}, "topics": [{"id": "https://openalex.org/T10883", "display_name": "Ethics and Social Impacts of AI", "score": 0.9785000085830688, "subfield": {"id": "https://openalex.org/subfields/3311", "display_name": "Safety Research"}, "field": {"id": "https://openalex.org/fields/33", "display_name": "Social Sciences"}, "domain": {"id": "https://openalex.org/domains/2", "display_name": "Social Sciences"}}, {"id": "https://openalex.org/T11689", "display_name": "Adversarial Robustness in Machine Learning", "score": 0.006300000008195639, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.0015999999595806003, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "doi:10.5281/zenodo.17526747", "is_oa": true, "landing_page_url": "https://doi.org/10.5281/zenodo.17526747", "pdf_url": null, "source": {"id": "https://openalex.org/S4306400562", "display_name": "Zenodo (CERN European Organization for Nuclear Research)", "issn_l": null, "issn": null, "is_oa": true, "is_in_doaj": false, "is_core": false, "listed_in": [], "host_organization": "https://openalex.org/I67311998", "host_organization_name": "European Organization for Nuclear Research", "host_organization_lineage": ["https://openalex.org/I67311998"], "host_organization_lineage_names": [], "type": "repository"}, "license": "cc-by", "license_id": "https://openalex.org/licenses/cc-by", "version": null, "is_accepted": false, "is_published": false, "raw_source_name": null, "raw_type": "article-journal"}, "cited_by_count": 0, "is_retracted": false}, {"id": "https://openalex.org/W4417357116", "doi": "https://doi.org/10.48550/arxiv.2506.10947", "title": "Spurious Rewards: Rethinking Training Signals in RLVR", "publication_year": 2025, "publication_date": "2025-06-12", "type": "preprint", "language": "en", "authorships": [{"author_position": "first", "author": {"id": "https://openalex.org/A5047742348", "display_name": "Rulin Shao", "orcid": "https://orcid.org/0000-0002-8228-2106", "observed_orcids": ["https://orcid.org/0000-0002-8228-2106"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Shao, Rulin", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5022934996", "display_name": "Shuyue Stella Li", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Li, Shuyue Stella", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5101888897", "display_name": "Rui Xin", "orcid": "https://orcid.org/0000-0002-0224-3431", "observed_orcids": ["https://orcid.org/0000-0002-0224-3431"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Xin, Rui", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5005776608", "display_name": "Scott Geng", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Geng, Scott", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5113844261", "display_name": "Yiping Wang", "orcid": "https://orcid.org/0000-0002-7540-6449", "observed_orcids": ["https://orcid.org/0000-0002-7540-6449"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Wang, Yiping", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5028243041", "display_name": "Sewoong Oh", "orcid": "https://orcid.org/0000-0002-8975-8306", "observed_orcids": ["https://orcid.org/0000-0002-8975-8306"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Oh, Sewoong", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5033061754", "display_name": "Simon S. Du", "orcid": "https://orcid.org/0000-0003-0056-8299", "observed_orcids": ["https://orcid.org/0000-0003-0056-8299"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Du, Simon Shaolei", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5083548887", "display_name": "Nathan Lambert", "orcid": "https://orcid.org/0000-0002-9997-6817", "observed_orcids": ["https://orcid.org/0000-0002-9997-6817"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Lambert, Nathan", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5039158419", "display_name": "Sewon Min", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Min, Sewon", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5032451496", "display_name": "Ranjay Krishna", "orcid": "https://orcid.org/0000-0001-8784-2531", "observed_orcids": ["https://orcid.org/0000-0001-8784-2531"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Krishna, Ranjay", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5062910836", "display_name": "Yulia Tsvetkov", "orcid": "https://orcid.org/0000-0002-4634-7128", "observed_orcids": ["https://orcid.org/0000-0002-4634-7128"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Tsvetkov, Yulia", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5082305994", "display_name": "Hannaneh Hajishirzi", "orcid": "https://orcid.org/0000-0002-1055-6657", "observed_orcids": ["https://orcid.org/0000-0002-1055-6657"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Hajishirzi, Hannaneh", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5079904764", "display_name": "Pang Wei Koh", "orcid": "https://orcid.org/0000-0003-4330-6969", "observed_orcids": ["https://orcid.org/0000-0003-4330-6969"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Koh, Pang Wei", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "last", "author": {"id": "https://openalex.org/A5067919401", "display_name": "Luke Zettlemoyer", "orcid": "https://orcid.org/0009-0008-8296-0764", "observed_orcids": ["https://orcid.org/0009-0008-8296-0764"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Zettlemoyer, Luke", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}], "abstract_inverted_index": {"We": [0], "show": [1, 67], "that": [2, 22, 68, 142, 194], "reinforcement": [3], "learning": [4], "with": [5, 19, 30, 38, 125], "verifiable": [6], "rewards": [7, 21, 141, 193], "(RLVR)": [8], "can": [9, 79, 185], "elicit": [10], "strong": [11], "mathematical": [12], "reasoning": [13, 106, 108], "in": [14, 99, 109], "certain": [15], "language": [16], "models": [17, 147, 173, 189], "even": [18, 27, 86, 190], "spurious": [20, 126, 140], "have": [23], "little,": [24], "no,": [25], "or": [26, 160], "negative": [28], "correlation": [29], "the": [31, 55, 75, 129, 165], "correct": [32], "answer.": [33], "For": [34], "example,": [35], "RLVR": [36], "training": [37], "GRPO": [39, 69], "improves": [40], "MATH-500": [41], "performance": [42], "for": [43, 145, 153], "Qwen2.5-Math-7B": [44], "by": [45], "21.4": [46], "percentage": [47], "points": [48], "using": [49], "randomly": [50], "assigned": [51], "rewards,": [52], "nearly": [53], "matching": [54], "29.1-point": [56], "gain": [57], "from": [58, 74, 118, 191], "ground-truth": [59], "rewards.": [60, 89, 127], "To": [61], "explain": [62], "this": [63], "counterintuitive": [64], "observation,": [65], "we": [66, 94, 103], "exhibits": [70], "a": [71, 91, 178], "clipping": [72], "bias": [73], "clip": [76], "term,": [77], "which": [78, 102], "amplify": [80], "high-prior": [81], "behaviors": [82, 134], "learned": [83], "during": [84], "pretraining": [85], "without": [87, 111], "informative": [88], "As": [90], "case": [92], "study,": [93], "identify": [95], "one": [96], "such": [97, 132, 157], "behavior": [98], "Qwen2.5-Math": [100], "models,": [101], "call": [104], "code": [105, 110, 113], "--": [107], "actual": [112], "execution;": [114], "code-reasoning": [115], "frequency": [116], "increases": [117], "65": [119], "percent": [120, 124], "to": [121, 150], "over": [122], "90": [123], "However,": [128], "presence": [130], "of": [131, 167], "amplifiable": [133], "is": [135], "highly": [136], "model-dependent.": [137], "In": [138], "practice,": [139], "are": [143], "effective": [144], "Qwen": [146, 188], "often": [148], "fail": [149], "produce": [151], "gains": [152, 184], "other": [154], "model": [155], "families,": [156], "as": [158], "Llama3": [159], "OLMo2.": [161], "Our": [162], "results": [163], "highlight": [164], "importance": [166], "validating": [168], "RL": [169], "methods": [170], "across": [171], "diverse": [172], "rather": [174], "than": [175], "relying": [176], "on": [177, 187], "single": [179], "de": [180], "facto": [181], "choice:": [182], "large": [183], "arise": [186], "random": [192], "do": [195], "not": [196], "reflect": [197], "genuine": [198], "capability": [199], "improvements.": [200]}, "primary_topic": {"id": "https://openalex.org/T10462", "display_name": "Reinforcement Learning in Robotics", "score": 0.4099000096321106, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T10462", "display_name": "Reinforcement Learning in Robotics", "score": 0.4099000096321106, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.11789999902248383, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10181", "display_name": "Natural Language Processing Techniques", "score": 0.04749999940395355, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "pmh:oai:arXiv.org:2506.10947", "is_oa": true, "landing_page_url": "http://arxiv.org/abs/2506.10947", "pdf_url": "https://arxiv.org/pdf/2506.10947", "source": {"id": "https://openalex.org/S4306400194", "display_name": "arXiv (Cornell University)", "issn_l": "2331-8422", "issn": ["2331-8422"], "is_oa": true, "is_in_doaj": false, "is_core": false, "listed_in": [], "host_organization": "https://openalex.org/I205783295", "host_organization_name": "Cornell University", "host_organization_lineage": ["https://openalex.org/I205783295"], "host_organization_lineage_names": [], "type": "repository"}, "license": null, "license_id": null, "version": "submittedVersion", "is_accepted": false, "is_published": false, "raw_source_name": null, "raw_type": "text"}, "cited_by_count": 0, "is_retracted": false}, {"id": "https://openalex.org/W4417098542", "doi": "https://doi.org/10.48550/arxiv.2508.01198", "title": "Adaptive Content Restriction for Large Language Models via Suffix Optimization", "publication_year": 2025, "publication_date": "2025-08-02", "type": "preprint", "language": "en", "authorships": [{"author_position": "first", "author": {"id": "https://openalex.org/A5101879293", "display_name": "Yige Li", "orcid": "https://orcid.org/0000-0002-9904-3611", "observed_orcids": ["https://orcid.org/0000-0002-9904-3611"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Li, Yige", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": null, "display_name": "Jiang, Peihai", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Jiang, Peihai", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5101729857", "display_name": "Jun Sun", "orcid": "https://orcid.org/0000-0002-4064-0315", "observed_orcids": ["https://orcid.org/0000-0002-4064-0315"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Sun, Jun", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5102713928", "display_name": "Peng Shu", "orcid": "https://orcid.org/0000-0002-5970-9992", "observed_orcids": ["https://orcid.org/0000-0002-5970-9992"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Shu, Peng", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5100647159", "display_name": "Tianming Liu", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Liu, Tianming", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "last", "author": {"id": "https://openalex.org/A5085283385", "display_name": "Zhen James Xiang", "orcid": "https://orcid.org/0000-0002-4284-2041", "observed_orcids": ["https://orcid.org/0000-0002-4284-2041"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Xiang, Zhen", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}], "abstract_inverted_index": {"Large": [0], "Language": [1], "Models": [2], "(LLMs)": [3], "have": [4], "demonstrated": [5], "significant": [6, 17], "success": [7], "across": [8, 55, 190], "diverse": [9], "applications.": [10], "However,": [11], "enforcing": [12], "content": [13, 28, 36, 50], "restrictions": [14], "remains": [15], "a": [16, 97, 141, 151, 156, 176], "challenge": [18], "due": [19, 82], "to": [20, 73, 83, 116, 145, 148], "their": [21], "expansive": [22], "output": [23, 165], "space.": [24], "One": [25], "aspect": [26], "of": [27, 69, 75, 158, 199], "restriction": [29, 51, 221], "is": [30, 80, 236], "preventing": [31], "LLMs": [32, 119], "from": [33, 120, 154], "generating": [34, 121, 155], "harmful": [35], "via": [37], "model": [38, 113], "alignment": [39], "approaches": [40], "such": [41, 208], "as": [42, 209], "supervised": [43], "fine-tuning": [44, 114], "(SFT).": [45], "Yet,": [46], "the": [47, 84, 130, 164, 197, 205], "need": [48], "for": [49, 124, 133, 186, 223], "may": [52], "vary": [53], "significantly": [54], "user": [56], "groups,": [57], "change": [58], "rapidly": [59], "over": [60], "time,": [61], "and": [62, 88, 217, 228], "not": [63], "always": [64], "align": [65], "with": [66], "general": [67], "definitions": [68], "harmfulness.": [70], "Applying": [71], "SFT": [72], "each": [74], "these": [76], "specific": [77, 125], "use": [78, 126], "cases": [79], "impractical": [81], "high": [85], "computational,": [86], "data,": [87], "storage": [89], "demands.": [90], "Motivated": [91], "by": [92, 212], "this": [93], "need,": [94], "we": [95, 174], "propose": [96, 129], "new": [98, 177], "task": [99], "called": [100], "\\textit{Adaptive": [101], "Content": [102], "Restriction}": [103], "(AdaCoRe),": [104], "which": [105, 139, 182, 203], "focuses": [106], "on": [107, 201, 219, 238], "lightweight": [108], "strategies": [109], "--": [110, 115], "methods": [111], "without": [112], "prevent": [117, 150], "deployed": [118], "restricted": [122, 159, 188], "terms": [123, 189], "cases.": [127], "We": [128, 195, 231], "first": [131], "method": [132], "AdaCoRe,": [134], "named": [135], "\\textit{Suffix": [136], "Optimization": [137], "(SOP)},": [138], "appends": [140], "short,": [142], "optimized": [143], "suffix": [144, 211], "any": [146], "prompt": [147], "a)": [149], "target": [152], "LLM": [153], "set": [157], "terms,": [160], "while": [161], "b)": [162], "preserving": [163], "quality.": [166], "To": [167], "evaluate": [168], "AdaCoRe": [169], "approaches,": [170], "including": [171], "our": [172], "SOP,": [173], "create": [175], "\\textit{Content": [178], "Restriction": [179], "Benchmark}": [180], "(CoReBench),": [181], "contains": [183], "400": [184], "prompts": [185], "80": [187], "8": [191], "carefully": [192], "selected": [193], "categories.": [194], "demonstrate": [196, 233], "effectiveness": [198], "SOP": [200, 235], "CoReBench,": [202], "outperforms": [204], "system-level": [206], "baselines": [207], "system": [210], "15\\%,": [213], "17\\%,": [214], "10\\%,": [215], "9\\%,": [216], "6\\%": [218], "average": [220], "rates": [222], "Gemma2-2B,": [224], "Mistral-7B,": [225], "Vicuna-7B,": [226], "Llama3-8B,": [227], "Llama3.1-8B,": [229], "respectively.": [230], "also": [232], "that": [234], "effective": [237], "POE,": [239], "an": [240], "online": [241], "platform": [242], "hosting": [243], "various": [244], "commercial": [245], "LLMs,": [246], "highlighting": [247], "its": [248], "practicality": [249], "in": [250], "real-world": [251], "scenarios.": [252]}, "primary_topic": {"id": "https://openalex.org/T11689", "display_name": "Adversarial Robustness in Machine Learning", "score": 0.43630000948905945, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T11689", "display_name": "Adversarial Robustness in Machine Learning", "score": 0.43630000948905945, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.2888000011444092, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T12262", "display_name": "Hate Speech and Cyberbullying Detection", "score": 0.18289999663829803, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "pmh:oai:arXiv.org:2508.01198", "is_oa": true, "landing_page_url": "http://arxiv.org/abs/2508.01198", "pdf_url": "https://arxiv.org/pdf/2508.01198", "source": {"id": "https://openalex.org/S4306400194", "display_name": "arXiv (Cornell University)", "issn_l": "2331-8422", "issn": ["2331-8422"], "is_oa": true, "is_in_doaj": false, "is_core": false, "listed_in": [], "host_organization": "https://openalex.org/I205783295", "host_organization_name": "Cornell University", "host_organization_lineage": ["https://openalex.org/I205783295"], "host_organization_lineage_names": [], "type": "repository"}, "license": null, "license_id": null, "version": "submittedVersion", "is_accepted": false, "is_published": false, "raw_source_name": null, "raw_type": "text"}, "cited_by_count": 0, "is_retracted": false}, {"id": "https://openalex.org/W4416191651", "doi": "https://doi.org/10.48550/arxiv.2511.08577", "title": "Think-at-Hard: Dynamic Looped Transformers for Improved Reasoning", "publication_year": 2025, "publication_date": "2025-11-11", "type": "preprint", "language": null, "authorships": [{"author_position": "first", "author": {"id": "https://openalex.org/A5101765987", "display_name": "Tianyu Fu", "orcid": "https://orcid.org/0000-0002-5441-9750", "observed_orcids": ["https://orcid.org/0000-0002-5441-9750"]}, "institutions": [{"id": "https://openalex.org/I99065089", "display_name": "Tsinghua University", "ror": "https://ror.org/03cve4549", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I99065089"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Fu, Tianyu", "raw_affiliation_strings": ["Infinigence AI", "Tsinghua University"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Infinigence AI", "institution_ids": []}, {"raw_affiliation_string": "Tsinghua University", "institution_ids": ["https://openalex.org/I99065089"]}]}, {"author_position": "middle", "author": {"id": null, "display_name": "You, Yichen", "orcid": null, "observed_orcids": []}, "institutions": [{"id": "https://openalex.org/I99065089", "display_name": "Tsinghua University", "ror": "https://ror.org/03cve4549", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I99065089"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "You, Yichen", "raw_affiliation_strings": ["Tsinghua University"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Tsinghua University", "institution_ids": ["https://openalex.org/I99065089"]}]}, {"author_position": "middle", "author": {"id": null, "display_name": "Chen, Zekai", "orcid": null, "observed_orcids": []}, "institutions": [{"id": "https://openalex.org/I99065089", "display_name": "Tsinghua University", "ror": "https://ror.org/03cve4549", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I99065089"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Chen, Zekai", "raw_affiliation_strings": ["Tsinghua University"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Tsinghua University", "institution_ids": ["https://openalex.org/I99065089"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5015946486", "display_name": "Guohao Dai", "orcid": "https://orcid.org/0000-0003-0849-3252", "observed_orcids": ["https://orcid.org/0000-0003-0849-3252", "https://orcid.org/0000-0002-3968-2980", "https://orcid.org/0000-0002-8464-0130"]}, "institutions": [{"id": "https://openalex.org/I183067930", "display_name": "Shanghai Jiao Tong University", "ror": "https://ror.org/0220qvk04", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I183067930"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Dai, Guohao", "raw_affiliation_strings": ["Infinigence AI", "Shanghai Jiao Tong University"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Infinigence AI", "institution_ids": []}, {"raw_affiliation_string": "Shanghai Jiao Tong University", "institution_ids": ["https://openalex.org/I183067930"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5023755254", "display_name": "Huazhong Yang", "orcid": "https://orcid.org/0000-0003-2421-353X", "observed_orcids": ["https://orcid.org/0000-0003-2421-353X"]}, "institutions": [{"id": "https://openalex.org/I99065089", "display_name": "Tsinghua University", "ror": "https://ror.org/03cve4549", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I99065089"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Yang, Huazhong", "raw_affiliation_strings": ["Tsinghua University"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Tsinghua University", "institution_ids": ["https://openalex.org/I99065089"]}]}, {"author_position": "last", "author": {"id": "https://openalex.org/A5100701610", "display_name": "Yu Wang", "orcid": "https://orcid.org/0000-0002-7216-6929", "observed_orcids": ["https://orcid.org/0000-0002-7216-6929"]}, "institutions": [{"id": "https://openalex.org/I99065089", "display_name": "Tsinghua University", "ror": "https://ror.org/03cve4549", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I99065089"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Wang, Yu", "raw_affiliation_strings": ["Tsinghua University"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Tsinghua University", "institution_ids": ["https://openalex.org/I99065089"]}]}], "abstract_inverted_index": {"Improving": [0], "the": [1, 50, 120, 133, 150, 214], "reasoning": [2], "abilities": [3], "of": [4, 196], "Large": [5], "Language": [6], "Models": [7], "(LLMs),": [8], "especially": [9], "under": [10], "parameter": [11, 183], "constraints,": [12], "is": [13, 225], "crucial": [14], "for": [15, 99], "real-world": [16], "applications.": [17], "Looped": [18], "transformers": [19], "address": [20], "this": [21], "by": [22, 84, 89, 189, 203], "performing": [23], "multiple": [24], "latent": [25, 40, 67, 110, 125], "iterations": [26, 68, 193], "to": [27, 86, 108, 116, 139, 154, 218], "refine": [28], "each": [29], "token": [30, 44, 151], "beyond": [31], "a": [32, 39, 95, 104], "single": [33], "forward": [34, 122], "pass.": [35, 123], "However,": [36], "we": [37, 91], "identify": [38], "overthinking": [41], "phenomenon:": [42], "most": [43], "predictions": [45], "are": [46, 54], "already": [47], "correct": [48], "after": [49, 119], "first": [51], "pass,": [52], "but": [53], "sometimes": [55], "revised": [56], "into": [57], "errors": [58], "in": [59], "later": [60], "iterations.": [61], "We": [62], "ask": [63], "whether": [64], "selectively": [65], "skipping": [66, 192], "can": [69], "improve": [70], "accuracy,": [71], "and": [72, 178, 198, 212, 220], "reveal": [73], "significant": [74], "potential": [75], "with": [76, 164], "an": [77, 155], "oracle": [78], "iteration": [79, 157], "policy": [80], "that": [81], "boosts": [82], "performance": [83], "up": [85], "7.3%.": [87], "Motivated": [88], "this,": [90], "propose": [92], "Think-at-Hard": [93], "(TaH),": [94], "looped": [96], "transformer": [97], "optimized": [98], "selective": [100], "iteration.": [101], "TaH": [102, 185], "employs": [103], "lightweight": [105], "neural": [106], "decider": [107], "trigger": [109], "iteration,": [111], "only": [112], "at": [113, 227], "tokens": [114], "likely": [115], "be": [117], "incorrect": [118], "standard": [121], "During": [124], "iterations,": [126], "depth-aware": [127], "Low-Rank": [128], "Adaptation": [129], "(LoRA)": [130], "modules": [131], "shift": [132], "objective": [134], "from": [135, 149, 210], "general": [136], "next-token": [137], "prediction": [138], "focused": [140], "hard-token": [141], "refinement.": [142], "A": [143], "duo-causal": [144], "attention": [145, 148], "mechanism": [146], "extends": [147], "sequence": [152], "dimension": [153], "additional": [156], "depth": [158], "dimension,": [159], "enabling": [160], "cross-iteration": [161], "information": [162], "flow": [163], "full": [165], "sequential": [166], "parallelism.": [167], "Experiments": [168], "on": [169, 194], "nine": [170], "benchmarks": [171], "show": [172], "consistent": [173], "gains": [174, 215], "across": [175], "math,": [176], "QA,": [177], "coding": [179], "tasks.": [180], "With": [181], "identical": [182], "counts,": [184], "outperforms": [186], "always-iterate": [187], "baselines": [188, 202], "3.8-4.4%": [190], "while": [191], "93%": [195], "tokens,": [197], "exceeds": [199], "single-iteration": [200], "Qwen3": [201], "3.0-3.8%.": [204], "When": [205], "allowing": [206], "<3%": [207], "more": [208], "parameters": [209], "LoRA": [211], "decider,": [213], "further": [216], "increase": [217], "5.3-6.2%": [219], "6.1-6.8%,": [221], "respectively.": [222], "Our": [223], "code": [224], "available": [226], "https://github.com/thu-nics/TaH.": [228]}, "primary_topic": {"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.9063000082969666, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.9063000082969666, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10181", "display_name": "Natural Language Processing Techniques", "score": 0.05660000070929527, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T11714", "display_name": "Multimodal Machine Learning Applications", "score": 0.012900000438094139, "subfield": {"id": "https://openalex.org/subfields/1707", "display_name": "Computer Vision and Pattern Recognition"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "pmh:oai:arXiv.org:2511.08577", "is_oa": true, "landing_page_url": "https://arxiv.org/abs/2511.08577", "pdf_url": "https://arxiv.org/pdf/2511.08577", "source": {"id": "https://openalex.org/S4306400194", "display_name": "arXiv (Cornell University)", "issn_l": "2331-8422", "issn": ["2331-8422"], "is_oa": true, "is_in_doaj": false, "is_core": false, "listed_in": [], "host_organization": "https://openalex.org/I205783295", "host_organization_name": "Cornell University", "host_organization_lineage": ["https://openalex.org/I205783295"], "host_organization_lineage_names": [], "type": "repository"}, "license": null, "license_id": null, "version": "submittedVersion", "is_accepted": false, "is_published": false, "raw_source_name": null, "raw_type": "text"}, "cited_by_count": 0, "is_retracted": false}, {"id": "https://openalex.org/W4415029110", "doi": "https://doi.org/10.48550/arxiv.2505.01903", "title": "LookAlike: Consistent Distractor Generation in Math MCQs", "publication_year": 2025, "publication_date": "2025-05-03", "type": "preprint", "language": "en", "authorships": [{"author_position": "first", "author": {"id": "https://openalex.org/A5113230478", "display_name": "Nisarg Parikh", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Parikh, Nisarg", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5073581485", "display_name": "Nigel Fernandez", "orcid": "https://orcid.org/0000-0002-2258-0208", "observed_orcids": ["https://orcid.org/0000-0002-2258-0208"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Fernandez, Nigel", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5008589669", "display_name": "Alexander Scarlatos", "orcid": "https://orcid.org/0000-0001-6419-5909", "observed_orcids": ["https://orcid.org/0000-0001-6419-5909"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Scarlatos, Alexander", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5034068210", "display_name": "Simon Woodhead", "orcid": "https://orcid.org/0000-0002-2192-9797", "observed_orcids": ["https://orcid.org/0000-0002-2192-9797"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Woodhead, Simon", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "last", "author": {"id": "https://openalex.org/A5063813962", "display_name": "Andrew Lan", "orcid": "https://orcid.org/0000-0002-8475-6600", "observed_orcids": ["https://orcid.org/0000-0002-8475-6600"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Lan, Andrew", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}], "abstract_inverted_index": {"Large": [0], "language": [1], "models": [2], "(LLMs)": [3], "are": [4, 23, 31], "increasingly": [5], "used": [6], "to": [7, 73], "generate": [8], "distractors": [9, 30, 151], "for": [10, 146], "multiple-choice": [11], "questions": [12], "(MCQs),": [13], "especially": [14], "in": [15, 25, 116, 121], "domains": [16], "like": [17], "math": [18, 110, 149], "education.": [19], "However,": [20], "existing": [21, 129], "approaches": [22], "limited": [24], "ensuring": [26], "that": [27, 42, 79], "the": [28, 138], "generated": [29], "consistent": [32, 148], "with": [33, 68], "common": [34], "student": [35], "errors.": [36], "We": [37], "propose": [38], "LookAlike,": [39], "a": [40, 105], "method": [41, 131], "improves": [43], "error-distractor": [44], "consistency": [45], "via": [46], "preference": [47, 57, 86], "optimization.": [48], "Our": [49], "two": [50], "main": [51], "innovations": [52], "are:": [53], "(a)": [54], "mining": [55, 145], "synthetic": [56], "pairs": [58], "from": [59], "model": [60], "inconsistencies,": [61], "and": [62, 100, 119, 143], "(b)": [63], "alternating": [64], "supervised": [65], "fine-tuning": [66], "(SFT)": [67], "Direct": [69], "Preference": [70], "Optimization": [71], "(DPO)": [72], "stabilize": [74], "training.": [75, 102], "Unlike": [76], "prior": [77], "work": [78], "relies": [80], "on": [81, 104], "heuristics": [82], "or": [83], "manually": [84], "annotated": [85], "data,": [87], "LookAlike": [88, 112], "uses": [89], "its": [90], "own": [91], "generation": [92, 118, 123], "inconsistencies": [93], "as": [94], "dispreferred": [95], "samples,": [96], "thus": [97], "enabling": [98], "scalable": [99], "stable": [101], "Evaluated": [103], "real-world": [106], "dataset": [107], "of": [108, 140], "1,400+": [109], "MCQs,": [111], "achieves": [113], "51.6%": [114], "accuracy": [115], "distractor": [117], "57.2%": [120], "error": [122], "under": [124], "LLM-as-a-judge": [125], "evaluation,": [126], "outperforming": [127], "an": [128], "state-of-the-art": [130], "(45.6%": [132], "/": [133], "47.7%).": [134], "These": [135], "improvements": [136], "highlight": [137], "effectiveness": [139], "preference-based": [141], "regularization": [142], "inconsistency": [144], "generating": [147], "MCQ": [150], "at": [152], "scale.": [153]}, "primary_topic": {"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.5429999828338623, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.5429999828338623, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T11902", "display_name": "Intelligent Tutoring Systems and Adaptive Learning", "score": 0.11299999803304672, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10181", "display_name": "Natural Language Processing Techniques", "score": 0.10740000009536743, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "pmh:oai:arXiv.org:2505.01903", "is_oa": true, "landing_page_url": "http://arxiv.org/abs/2505.01903", "pdf_url": "https://arxiv.org/pdf/2505.01903", "source": {"id": "https://openalex.org/S4306400194", "display_name": "arXiv (Cornell University)", "issn_l": "2331-8422", "issn": ["2331-8422"], "is_oa": true, "is_in_doaj": false, "is_core": false, "listed_in": [], "host_organization": "https://openalex.org/I205783295", "host_organization_name": "Cornell University", "host_organization_lineage": ["https://openalex.org/I205783295"], "host_organization_lineage_names": [], "type": "repository"}, "license": null, "license_id": null, "version": "submittedVersion", "is_accepted": false, "is_published": false, "raw_source_name": null, "raw_type": "text"}, "cited_by_count": 0, "is_retracted": false}, {"id": "https://openalex.org/W4417296930", "doi": "https://doi.org/10.48550/arxiv.2512.10756", "title": "OPV: Outcome-based Process Verifier for Efficient Long Chain-of-Thought Verification", "publication_year": 2025, "publication_date": "2025-12-11", "type": "preprint", "language": null, "authorships": [{"author_position": "first", "author": {"id": "https://openalex.org/A5075965565", "display_name": "Zijian Wu", "orcid": "https://orcid.org/0000-0002-6675-8231", "observed_orcids": ["https://orcid.org/0000-0002-6675-8231"]}, "institutions": [{"id": "https://openalex.org/I177725633", "display_name": "Chinese University of Hong Kong", "ror": "https://ror.org/00t33hh48", "country_code": "HK", "type": "education", "lineage": ["https://openalex.org/I177725633"]}, {"id": "https://openalex.org/I4391012619", "display_name": "Shanghai Artificial Intelligence Laboratory", "ror": "https://ror.org/03wkvpx79", "country_code": "CN", "type": "facility", "lineage": ["https://openalex.org/I4391012619"]}], "countries": ["CN", "HK"], "is_corresponding": false, "raw_author_name": "Wu, Zijian", "raw_affiliation_strings": ["MMLab , The Chinese University of Hong Kong ,", "Shanghai AI Laboratory ,"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "MMLab , The Chinese University of Hong Kong ,", "institution_ids": ["https://openalex.org/I177725633"]}, {"raw_affiliation_string": "Shanghai AI Laboratory ,", "institution_ids": ["https://openalex.org/I4391012619"]}]}, {"author_position": "middle", "author": {"id": null, "display_name": "Kong, Lingkai", "orcid": null, "observed_orcids": []}, "institutions": [{"id": "https://openalex.org/I183067930", "display_name": "Shanghai Jiao Tong University", "ror": "https://ror.org/0220qvk04", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I183067930"]}, {"id": "https://openalex.org/I4391012619", "display_name": "Shanghai Artificial Intelligence Laboratory", "ror": "https://ror.org/03wkvpx79", "country_code": "CN", "type": "facility", "lineage": ["https://openalex.org/I4391012619"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Kong, Lingkai", "raw_affiliation_strings": ["Shanghai AI Laboratory ,", "Shanghai Jiao Tong University"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Shanghai AI Laboratory ,", "institution_ids": ["https://openalex.org/I4391012619"]}, {"raw_affiliation_string": "Shanghai Jiao Tong University", "institution_ids": ["https://openalex.org/I183067930"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5100709484", "display_name": "Wenwei Zhang", "orcid": "https://orcid.org/0000-0002-2748-4514", "observed_orcids": ["https://orcid.org/0000-0002-2748-4514"]}, "institutions": [{"id": "https://openalex.org/I4391012619", "display_name": "Shanghai Artificial Intelligence Laboratory", "ror": "https://ror.org/03wkvpx79", "country_code": "CN", "type": "facility", "lineage": ["https://openalex.org/I4391012619"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Zhang, Wenwei", "raw_affiliation_strings": ["Shanghai AI Laboratory ,"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Shanghai AI Laboratory ,", "institution_ids": ["https://openalex.org/I4391012619"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5109932574", "display_name": "Songyang Gao", "orcid": "https://orcid.org/0000-0001-5485-8019", "observed_orcids": ["https://orcid.org/0000-0001-5485-8019"]}, "institutions": [{"id": "https://openalex.org/I4391012619", "display_name": "Shanghai Artificial Intelligence Laboratory", "ror": "https://ror.org/03wkvpx79", "country_code": "CN", "type": "facility", "lineage": ["https://openalex.org/I4391012619"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Gao, Songyang", "raw_affiliation_strings": ["Shanghai AI Laboratory ,"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Shanghai AI Laboratory ,", "institution_ids": ["https://openalex.org/I4391012619"]}]}, {"author_position": "middle", "author": {"id": null, "display_name": "Gu, Yuzhe", "orcid": null, "observed_orcids": []}, "institutions": [{"id": "https://openalex.org/I4391012619", "display_name": "Shanghai Artificial Intelligence Laboratory", "ror": "https://ror.org/03wkvpx79", "country_code": "CN", "type": "facility", "lineage": ["https://openalex.org/I4391012619"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Gu, Yuzhe", "raw_affiliation_strings": ["Shanghai AI Laboratory ,"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Shanghai AI Laboratory ,", "institution_ids": ["https://openalex.org/I4391012619"]}]}, {"author_position": "middle", "author": {"id": null, "display_name": "Cai, Zhongrui", "orcid": null, "observed_orcids": []}, "institutions": [{"id": "https://openalex.org/I4391012619", "display_name": "Shanghai Artificial Intelligence Laboratory", "ror": "https://ror.org/03wkvpx79", "country_code": "CN", "type": "facility", "lineage": ["https://openalex.org/I4391012619"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Cai, Zhongrui", "raw_affiliation_strings": ["Shanghai AI Laboratory ,"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Shanghai AI Laboratory ,", "institution_ids": ["https://openalex.org/I4391012619"]}]}, {"author_position": "middle", "author": {"id": null, "display_name": "Ma, Tianyou", "orcid": null, "observed_orcids": []}, "institutions": [{"id": "https://openalex.org/I4391012619", "display_name": "Shanghai Artificial Intelligence Laboratory", "ror": "https://ror.org/03wkvpx79", "country_code": "CN", "type": "facility", "lineage": ["https://openalex.org/I4391012619"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Ma, Tianyou", "raw_affiliation_strings": ["Shanghai AI Laboratory ,"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Shanghai AI Laboratory ,", "institution_ids": ["https://openalex.org/I4391012619"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5100324598", "display_name": "Yuhong Liu", "orcid": "https://orcid.org/0000-0001-9364-0039", "observed_orcids": ["https://orcid.org/0000-0001-9364-0039"]}, "institutions": [{"id": "https://openalex.org/I4391012619", "display_name": "Shanghai Artificial Intelligence Laboratory", "ror": "https://ror.org/03wkvpx79", "country_code": "CN", "type": "facility", "lineage": ["https://openalex.org/I4391012619"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Liu, Yuhong", "raw_affiliation_strings": ["Shanghai AI Laboratory ,"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Shanghai AI Laboratory ,", "institution_ids": ["https://openalex.org/I4391012619"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5025948413", "display_name": "Zhi Wang", "orcid": "https://orcid.org/0000-0003-3398-9954", "observed_orcids": ["https://orcid.org/0000-0003-3398-9954", "https://orcid.org/0000-0003-4631-9669"]}, "institutions": [{"id": "https://openalex.org/I4391012619", "display_name": "Shanghai Artificial Intelligence Laboratory", "ror": "https://ror.org/03wkvpx79", "country_code": "CN", "type": "facility", "lineage": ["https://openalex.org/I4391012619"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Wang, Zhi", "raw_affiliation_strings": ["Shanghai AI Laboratory ,"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Shanghai AI Laboratory ,", "institution_ids": ["https://openalex.org/I4391012619"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5007188271", "display_name": "Runyuan Ma", "orcid": null, "observed_orcids": []}, "institutions": [{"id": "https://openalex.org/I4391012619", "display_name": "Shanghai Artificial Intelligence Laboratory", "ror": "https://ror.org/03wkvpx79", "country_code": "CN", "type": "facility", "lineage": ["https://openalex.org/I4391012619"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Ma, Runyuan", "raw_affiliation_strings": ["Shanghai AI Laboratory ,"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Shanghai AI Laboratory ,", "institution_ids": ["https://openalex.org/I4391012619"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5100458579", "display_name": "Guangyu Wang", "orcid": "https://orcid.org/0000-0003-4803-7200", "observed_orcids": ["https://orcid.org/0000-0003-4803-7200"]}, "institutions": [{"id": "https://openalex.org/I4391012619", "display_name": "Shanghai Artificial Intelligence Laboratory", "ror": "https://ror.org/03wkvpx79", "country_code": "CN", "type": "facility", "lineage": ["https://openalex.org/I4391012619"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Wang, Guangyu", "raw_affiliation_strings": ["Shanghai AI Laboratory ,"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Shanghai AI Laboratory ,", "institution_ids": ["https://openalex.org/I4391012619"]}]}, {"author_position": "middle", "author": {"id": null, "display_name": "Li, Wei", "orcid": null, "observed_orcids": []}, "institutions": [{"id": "https://openalex.org/I4391012619", "display_name": "Shanghai Artificial Intelligence Laboratory", "ror": "https://ror.org/03wkvpx79", "country_code": "CN", "type": "facility", "lineage": ["https://openalex.org/I4391012619"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Li, Wei", "raw_affiliation_strings": ["Shanghai AI Laboratory ,"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Shanghai AI Laboratory ,", "institution_ids": ["https://openalex.org/I4391012619"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5101615091", "display_name": "Conghui He", "orcid": "https://orcid.org/0000-0001-8697-695X", "observed_orcids": ["https://orcid.org/0000-0001-8697-695X"]}, "institutions": [{"id": "https://openalex.org/I4391012619", "display_name": "Shanghai Artificial Intelligence Laboratory", "ror": "https://ror.org/03wkvpx79", "country_code": "CN", "type": "facility", "lineage": ["https://openalex.org/I4391012619"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "He, Conghui", "raw_affiliation_strings": ["Shanghai AI Laboratory ,"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Shanghai AI Laboratory ,", "institution_ids": ["https://openalex.org/I4391012619"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5010087030", "display_name": "Dahua Lin", "orcid": "https://orcid.org/0000-0002-8865-7896", "observed_orcids": ["https://orcid.org/0000-0002-8865-7896"]}, "institutions": [{"id": "https://openalex.org/I177725633", "display_name": "Chinese University of Hong Kong", "ror": "https://ror.org/00t33hh48", "country_code": "HK", "type": "education", "lineage": ["https://openalex.org/I177725633"]}, {"id": "https://openalex.org/I4391012619", "display_name": "Shanghai Artificial Intelligence Laboratory", "ror": "https://ror.org/03wkvpx79", "country_code": "CN", "type": "facility", "lineage": ["https://openalex.org/I4391012619"]}], "countries": ["CN", "HK"], "is_corresponding": false, "raw_author_name": "Lin, Dahua", "raw_affiliation_strings": ["MMLab , The Chinese University of Hong Kong ,", "Shanghai AI Laboratory ,"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "MMLab , The Chinese University of Hong Kong ,", "institution_ids": ["https://openalex.org/I177725633"]}, {"raw_affiliation_string": "Shanghai AI Laboratory ,", "institution_ids": ["https://openalex.org/I4391012619"]}]}, {"author_position": "last", "author": {"id": "https://openalex.org/A5100437926", "display_name": "Kai Chen", "orcid": "https://orcid.org/0000-0001-6850-1723", "observed_orcids": ["https://orcid.org/0000-0001-6850-1723", "https://orcid.org/0000-0002-7665-7506", "https://orcid.org/0009-0000-7636-5848", "https://orcid.org/0000-0002-7233-5892"]}, "institutions": [{"id": "https://openalex.org/I4391012619", "display_name": "Shanghai Artificial Intelligence Laboratory", "ror": "https://ror.org/03wkvpx79", "country_code": "CN", "type": "facility", "lineage": ["https://openalex.org/I4391012619"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Chen, Kai", "raw_affiliation_strings": ["Shanghai AI Laboratory ,"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Shanghai AI Laboratory ,", "institution_ids": ["https://openalex.org/I4391012619"]}]}], "abstract_inverted_index": {"Large": [0], "language": [1], "models": [2, 197], "(LLMs)": [3], "have": [4, 58], "achieved": [5], "significant": [6], "progress": [7], "in": [8, 45, 60, 64, 142], "solving": [9], "complex": [10, 66], "reasoning": [11, 48], "tasks": [12], "by": [13, 29, 70], "Reinforcement": [14], "Learning": [15], "with": [16, 126, 137, 201, 221, 226], "Verifiable": [17], "Rewards": [18], "(RLVR).": [19], "This": [20], "advancement": [21], "is": [22], "also": [23], "inseparable": [24], "from": [25, 100, 240], "the": [26, 41, 46, 65, 71, 78, 87, 94, 116, 132, 145, 150, 172, 236, 247], "oversight": [27], "automated": [28], "reliable": [30], "verifiers.": [31], "However,": [32], "current": [33, 54, 151], "outcome-based": [34], "verifiers": [35, 56], "(OVs)": [36], "are": [37, 154], "unable": [38], "to": [39, 77, 103, 129, 160, 208, 242], "inspect": [40], "unreliable": [42], "intermediate": [43], "steps": [44], "long": [47, 67, 101], "chains": [49], "of": [50, 73, 81, 97, 135, 149, 205, 238], "thought": [51], "(CoTs).": [52], "Meanwhile,": [53], "process-based": [55], "(PVs)": [57], "difficulties": [59], "reliably": [61], "detecting": [62], "errors": [63], "CoTs,": [68], "limited": [69], "scarcity": [72], "high-quality": [74], "annotations": [75, 128], "due": [76], "prohibitive": [79], "costs": [80], "human": [82], "annotations.": [83], "Therefore,": [84], "we": [85, 119], "propose": [86], "Outcome-based": [88], "Process": [89], "Verifier": [90], "(OPV),": [91], "which": [92], "verifies": [93], "rationale": [95], "process": [96], "summarized": [98], "outcomes": [99], "CoTs": [102], "achieve": [104], "both": [105], "accurate": [106], "and": [107, 110, 156, 169, 181], "efficient": [108], "verification": [109, 133], "enable": [111], "large-scale": [112], "annotation.": [113], "To": [114], "empower": [115], "proposed": [117], "verifier,": [118], "adopt": [120], "an": [121, 202], "iterative": [122], "active": [123], "learning": [124], "framework": [125], "expert": [127, 222], "progressively": [130], "improve": [131], "capability": [134], "OPV": [136, 153, 164, 211, 229], "fewer": [138], "annotation": [139], "costs.": [140], "Specifically,": [141], "each": [143], "iteration,": [144], "most": [146], "uncertain": [147], "cases": [148], "best": [152], "annotated": [155], "then": [157], "subsequently": [158], "used": [159], "train": [161], "a": [162], "new": [163, 186], "through": [165], "Rejection": [166], "Fine-Tuning": [167], "(RFT)": [168], "RLVR": [170], "for": [171], "next": [173], "round.": [174], "Extensive": [175], "experiments": [176], "demonstrate": [177], "OPV's": [178], "superior": [179], "performance": [180, 232], "broad": [182], "applicability.": [183], "It": [184], "achieves": [185], "state-of-the-art": [187], "results": [188], "on": [189, 244], "our": [190], "held-out": [191], "OPV-Bench,": [192], "outperforming": [193], "much": [194], "larger": [195], "open-source": [196], "such": [198], "as": [199, 246], "Qwen3-Max-Preview": [200], "F1": [203], "score": [204], "83.1": [206], "compared": [207], "76.3.": [209], "Furthermore,": [210], "effectively": [212], "detects": [213], "false": [214], "positives": [215], "within": [216], "synthetic": [217], "dataset,": [218], "closely": [219], "align": [220], "assessment.": [223], "When": [224], "collaborating": [225], "policy": [227], "models,": [228], "consistently": [230], "yields": [231], "gains,": [233], "e.g.,": [234], "raising": [235], "accuracy": [237], "DeepSeek-R1-Distill-Qwen-32B": [239], "55.2%": [241], "73.3%": [243], "AIME2025": [245], "compute": [248], "budget": [249], "scales.": [250]}, "primary_topic": {"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.6098999977111816, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.6098999977111816, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10181", "display_name": "Natural Language Processing Techniques", "score": 0.15160000324249268, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T11689", "display_name": "Adversarial Robustness in Machine Learning", "score": 0.06159999966621399, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "pmh:oai:arXiv.org:2512.10756", "is_oa": true, "landing_page_url": "http://arxiv.org/abs/2512.10756", "pdf_url": "https://arxiv.org/pdf/2512.10756", "source": {"id": "https://openalex.org/S4306400194", "display_name": "arXiv (Cornell University)", "issn_l": "2331-8422", "issn": ["2331-8422"], "is_oa": true, "is_in_doaj": false, "is_core": false, "listed_in": [], "host_organization": "https://openalex.org/I205783295", "host_organization_name": "Cornell University", "host_organization_lineage": ["https://openalex.org/I205783295"], "host_organization_lineage_names": [], "type": "repository"}, "license": "cc-by", "license_id": "https://openalex.org/licenses/cc-by", "version": "submittedVersion", "is_accepted": false, "is_published": false, "raw_source_name": null, "raw_type": "text"}, "cited_by_count": 0, "is_retracted": false}, {"id": "https://openalex.org/W4416113701", "doi": "https://doi.org/10.48550/arxiv.2503.04592", "title": "A Benchmark for Multi-Lingual Vision-Language Learning in Remote Sensing Image Captioning", "publication_year": 2025, "publication_date": "2025-03-06", "type": "preprint", "language": "en", "authorships": [{"author_position": "first", "author": {"id": "https://openalex.org/A5004509724", "display_name": "Qing Zhou", "orcid": "https://orcid.org/0000-0002-2100-8840", "observed_orcids": ["https://orcid.org/0000-0002-2100-8840"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Zhou, Qing", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5060493155", "display_name": "Yang Tao", "orcid": "https://orcid.org/0000-0002-9535-464X", "observed_orcids": ["https://orcid.org/0000-0002-9535-464X"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Yang, Tao", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5014526931", "display_name": "Junyu Gao", "orcid": "https://orcid.org/0000-0002-8105-5497", "observed_orcids": ["https://orcid.org/0000-0002-8105-5497"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Gao, Junyu", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5102013700", "display_name": "Weiping Ni", "orcid": "https://orcid.org/0000-0002-0640-3922", "observed_orcids": ["https://orcid.org/0000-0002-0640-3922"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Ni, Weiping", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5042729155", "display_name": "Junzheng Wu", "orcid": "https://orcid.org/0000-0002-1267-3523", "observed_orcids": ["https://orcid.org/0000-0002-1267-3523"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Wu, Junzheng", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "last", "author": {"id": "https://openalex.org/A5100341321", "display_name": "Qi Wang", "orcid": "https://orcid.org/0000-0002-7028-4956", "observed_orcids": ["https://orcid.org/0000-0002-7028-4956", "https://orcid.org/0000-0002-6269-0196", "https://orcid.org/0009-0005-8281-0019"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Wang, Qi", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}], "abstract_inverted_index": {"Remote": [0, 100], "Sensing": [1, 101], "Image": [2, 102], "Captioning": [3], "(RSIC)": [4], "is": [5], "a": [6, 104, 132], "cross-modal": [7], "field": [8], "bridging": [9], "vision": [10], "and": [11, 22, 35, 53, 68, 96, 181, 193, 214], "language,": [12], "aimed": [13], "at": [14, 219], "automatically": [15], "generating": [16], "natural": [17], "language": [18], "descriptions": [19], "of": [20, 49, 56, 71, 77, 147, 163, 195], "features": [21], "scenes": [23], "in": [24, 31, 74, 141, 198], "remote": [25, 201], "sensing": [26, 202], "imagery.": [27], "Despite": [28], "significant": [29, 88], "advances": [30], "developing": [32], "sophisticated": [33], "methods": [34], "large-scale": [36], "datasets": [37, 52, 114], "for": [38, 60], "training": [39], "vision-language": [40, 167], "models": [41, 168], "(VLMs),": [42], "two": [43], "critical": [44], "challenges": [45], "persist:": [46], "the": [47, 54, 66, 75, 91, 138, 191], "scarcity": [48], "non-English": [50], "descriptive": [51], "lack": [55], "multilingual": [57, 200], "capability": [58], "evaluation": [59, 134, 142, 186], "models.": [61], "These": [62], "limitations": [63, 194], "fundamentally": [64], "impede": [65], "progress": [67], "practical": [69], "deployment": [70], "RSIC,": [72], "particularly": [73], "era": [76], "large": [78, 166], "VLMs.": [79], "To": [80], "address": [81], "these": [82], "challenges,": [83], "this": [84, 128], "paper": [85], "presents": [86], "several": [87], "contributions": [89], "to": [90], "field.": [92], "First,": [93], "we": [94, 130, 157], "introduce": [95], "analyze": [97], "BRSIC": [98], "(Bilingual": [99], "Captioning),": [103], "comprehensive": [105, 185], "bilingual": [106, 124], "dataset": [107], "that": [108, 136], "enriches": [109], "three": [110], "established": [111], "English": [112], "RSIC": [113], "with": [115, 122], "Chinese": [116], "descriptions,": [117], "encompassing": [118], "13,634": [119], "images": [120], "paired": [121], "68,170": [123], "captions.": [125], "Building": [126], "upon": [127], "foundation,": [129], "develop": [131], "systematic": [133], "framework": [135], "addresses": [137], "prevalent": [139], "inconsistency": [140], "protocols,": [143], "enabling": [144], "rigorous": [145], "assessment": [146], "model": [148], "performance": [149], "through": [150], "standardized": [151], "retraining": [152], "procedures": [153], "on": [154], "BRSIC.": [155], "Furthermore,": [156], "present": [158], "an": [159], "extensive": [160], "empirical": [161], "study": [162], "eight": [164], "state-of-the-art": [165], "(LVLMs),": [169], "examining": [170], "their": [171], "capabilities": [172], "across": [173], "multiple": [174], "paradigms": [175], "including": [176], "zero-shot": [177], "inference,": [178], "supervised": [179], "fine-tuning,": [180], "multi-lingual": [182], "training.": [183], "This": [184], "provides": [187], "crucial": [188], "insights": [189], "into": [190], "strengths": [192], "current": [196], "LVLMs": [197], "handling": [199], "tasks.": [203], "Additionally,": [204], "our": [205], "cross-dataset": [206], "transfer": [207], "experiments": [208], "reveal": [209], "interesting": [210], "findings.": [211], "The": [212], "code": [213], "data": [215], "will": [216], "be": [217], "available": [218], "https://github.com/mrazhou/BRSIC.": [220]}, "primary_topic": {"id": "https://openalex.org/T11714", "display_name": "Multimodal Machine Learning Applications", "score": 0.9990000128746033, "subfield": {"id": "https://openalex.org/subfields/1707", "display_name": "Computer Vision and Pattern Recognition"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T11714", "display_name": "Multimodal Machine Learning Applications", "score": 0.9990000128746033, "subfield": {"id": "https://openalex.org/subfields/1707", "display_name": "Computer Vision and Pattern Recognition"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10689", "display_name": "Remote-Sensing Image Classification", "score": 0.00019999999494757503, "subfield": {"id": "https://openalex.org/subfields/2214", "display_name": "Media Technology"}, "field": {"id": "https://openalex.org/fields/22", "display_name": "Engineering"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10036", "display_name": "Advanced Neural Network Applications", "score": 9.999999747378752e-05, "subfield": {"id": "https://openalex.org/subfields/1707", "display_name": "Computer Vision and Pattern Recognition"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "pmh:oai:arXiv.org:2503.04592", "is_oa": true, "landing_page_url": "http://arxiv.org/abs/2503.04592", "pdf_url": "https://arxiv.org/pdf/2503.04592", "source": {"id": "https://openalex.org/S4306400194", "display_name": "arXiv (Cornell University)", "issn_l": "2331-8422", "issn": ["2331-8422"], "is_oa": true, "is_in_doaj": false, "is_core": false, "listed_in": [], "host_organization": "https://openalex.org/I205783295", "host_organization_name": "Cornell University", "host_organization_lineage": ["https://openalex.org/I205783295"], "host_organization_lineage_names": [], "type": "repository"}, "license": null, "license_id": null, "version": "submittedVersion", "is_accepted": false, "is_published": false, "raw_source_name": null, "raw_type": "text"}, "cited_by_count": 0, "is_retracted": false}, {"id": "https://openalex.org/W4417281233", "doi": "https://doi.org/10.48550/arxiv.2503.10883", "title": "Chat-TS: Enhancing Multi-Modal Reasoning Over Time-Series and Natural Language Data", "publication_year": 2025, "publication_date": "2025-03-13", "type": "preprint", "language": "en", "authorships": [{"author_position": "first", "author": {"id": "https://openalex.org/A5112930614", "display_name": "Paul Quinlan", "orcid": "https://orcid.org/0000-0002-3273-4557", "observed_orcids": ["https://orcid.org/0000-0002-3273-4557"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Quinlan, Paul", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5100611696", "display_name": "Qingguo Li", "orcid": "https://orcid.org/0000-0001-8200-6550", "observed_orcids": ["https://orcid.org/0000-0001-8200-6550"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Li, Qingguo", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "last", "author": {"id": "https://openalex.org/A5016892586", "display_name": "Xiaodan Zhu", "orcid": "https://orcid.org/0000-0003-3856-3696", "observed_orcids": ["https://orcid.org/0000-0003-3856-3696", "https://orcid.org/0000-0002-9546-3176"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Zhu, Xiaodan", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}], "abstract_inverted_index": {"Large": [0], "language": [1, 53, 89, 191], "models": [2], "are": [3, 20, 26], "being": [4], "rapidly": [5], "deployed": [6], "across": [7], "many": [8], "fields": [9], "such": [10], "as": [11], "healthcare,": [12], "finance,": [13], "transportation,": [14], "and": [15, 39, 65, 94, 113, 122, 133, 151], "energy,": [16], "where": [17], "time-series": [18, 38, 73, 107, 174, 195], "data": [19, 108], "fundamental": [21], "components.": [22], "The": [23], "current": [24], "works": [25], "still": [27], "limited": [28], "in": [29, 183], "their": [30], "ability": [31, 81], "to": [32, 59, 129, 162], "perform": [33], "reasoning": [34, 61, 80, 147, 166, 185], "that": [35, 178], "involves": [36], "both": [37, 83], "the": [40, 100, 118, 164], "corresponding": [41], "textual": [42, 66], "content.": [43], "We": [44, 157], "address": [45], "this": [46], "gap": [47], "by": [48, 187], "introducing": [49], "Chat-TS,": [50], "a": [51, 134, 159], "large": [52], "model": [54], "(LLM)": [55], "based": [56], "framework": [57], "designed": [58], "support": [60, 92], "over": [62, 82], "time": [63], "series": [64], "data.": [67], "Unlike": [68], "traditional": [69], "models,": [70], "Chat-TS": [71, 179], "integrates": [72], "tokens": [74], "into": [75], "LLMs'": [76], "vocabulary,": [77], "enhancing": [78], "its": [79], "modalities": [84], "without": [85], "compromising": [86], "core": [87], "natural": [88, 190], "capabilities.": [90], "To": [91], "learning": [93], "evaluation,": [95], "we": [96], "contribute": [97], "new": [98], "datasets:": [99], "TS": [101, 119, 135, 144], "Instruct": [102, 120, 136, 145], "Training": [103], "Dataset": [104, 126], "(pairing": [105], "diverse": [106], "with": [109], "relevant": [110], "text": [111], "instructions": [112], "responses": [114], "for": [115, 154, 173], "instruction": [116], "tuning),": [117], "Question": [121], "Answer": [123], "(QA)": [124], "Gold": [125], "(multiple-choice": [127], "questions": [128, 153], "evaluate": [130], "multimodal": [131, 184], "reasoning),": [132], "Quantitative": [137], "Probing": [138], "Set": [139], "(a": [140], "small": [141], "subset": [142], "of": [143, 168], "QA": [146], "tasks": [148, 186], "alongside": [149], "math": [150], "decision-making": [152], "LLM": [155], "evaluation).": [156], "design": [158], "training": [160], "strategy": [161], "preserve": [163], "inherent": [165], "capabilities": [167], "LLMs": [169], "while": [170, 193], "augmenting": [171], "them": [172], "reasoning.": [175, 196], "Experiments": [176], "show": [177], "achieves": [180], "state-of-the-art": [181], "performance": [182], "maintaining": [188], "strong": [189], "proficiency": [192], "improving": [194]}, "primary_topic": {"id": "https://openalex.org/T11714", "display_name": "Multimodal Machine Learning Applications", "score": 0.9437999725341797, "subfield": {"id": "https://openalex.org/subfields/1707", "display_name": "Computer Vision and Pattern Recognition"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T11714", "display_name": "Multimodal Machine Learning Applications", "score": 0.9437999725341797, "subfield": {"id": "https://openalex.org/subfields/1707", "display_name": "Computer Vision and Pattern Recognition"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T12205", "display_name": "Time Series Analysis and Forecasting", "score": 0.041200000792741776, "subfield": {"id": "https://openalex.org/subfields/1711", "display_name": "Signal Processing"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.01209999993443489, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "pmh:oai:arXiv.org:2503.10883", "is_oa": true, "landing_page_url": "http://arxiv.org/abs/2503.10883", "pdf_url": "https://arxiv.org/pdf/2503.10883", "source": {"id": "https://openalex.org/S4306400194", "display_name": "arXiv (Cornell University)", "issn_l": "2331-8422", "issn": ["2331-8422"], "is_oa": true, "is_in_doaj": false, "is_core": false, "listed_in": [], "host_organization": "https://openalex.org/I205783295", "host_organization_name": "Cornell University", "host_organization_lineage": ["https://openalex.org/I205783295"], "host_organization_lineage_names": [], "type": "repository"}, "license": null, "license_id": null, "version": "submittedVersion", "is_accepted": false, "is_published": false, "raw_source_name": null, "raw_type": "text"}, "cited_by_count": 0, "is_retracted": false}, {"id": "https://openalex.org/W4413245533", "doi": "https://doi.org/10.1007/s13042-025-02760-4", "title": "Text classification methods in the steel e-commerce industry using the large language model and prompt engineering", "publication_year": 2025, "publication_date": "2025-08-08", "type": "article", "language": "en", "authorships": [{"author_position": "first", "author": {"id": "https://openalex.org/A5043859965", "display_name": "Qiaojuan Peng", "orcid": "https://orcid.org/0009-0008-8667-1468", "observed_orcids": ["https://orcid.org/0009-0008-8667-1468"]}, "institutions": [{"id": "https://openalex.org/I92403157", "display_name": "University of Science and Technology Beijing", "ror": "https://ror.org/02egmk993", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I92403157"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Qiaojuan Peng", "raw_affiliation_strings": ["Beijing Key Laboratory of Knowledge Engineering for Materials Science, Beijing, 100083, China", "School of Computer and Communication Engineering, University of Science and Technology Beijing, Beijing, 100083, China", "Shunde Innovation School, University of Science and Technology Beijing, Foshan, 528399, China"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Beijing Key Laboratory of Knowledge Engineering for Materials Science, Beijing, 100083, China", "institution_ids": []}, {"raw_affiliation_string": "School of Computer and Communication Engineering, University of Science and Technology Beijing, Beijing, 100083, China", "institution_ids": ["https://openalex.org/I92403157"]}, {"raw_affiliation_string": "Shunde Innovation School, University of Science and Technology Beijing, Foshan, 528399, China", "institution_ids": ["https://openalex.org/I92403157"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5070952629", "display_name": "Xiong Luo", "orcid": "https://orcid.org/0000-0002-1929-8447", "observed_orcids": ["https://orcid.org/0000-0002-1929-8447"]}, "institutions": [{"id": "https://openalex.org/I92403157", "display_name": "University of Science and Technology Beijing", "ror": "https://ror.org/02egmk993", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I92403157"]}], "countries": ["CN"], "is_corresponding": true, "raw_author_name": "Xiong Luo", "raw_affiliation_strings": ["Beijing Key Laboratory of Knowledge Engineering for Materials Science, Beijing, 100083, China", "School of Computer and Communication Engineering, University of Science and Technology Beijing, Beijing, 100083, China", "Shunde Innovation School, University of Science and Technology Beijing, Foshan, 528399, China"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Beijing Key Laboratory of Knowledge Engineering for Materials Science, Beijing, 100083, China", "institution_ids": []}, {"raw_affiliation_string": "School of Computer and Communication Engineering, University of Science and Technology Beijing, Beijing, 100083, China", "institution_ids": ["https://openalex.org/I92403157"]}, {"raw_affiliation_string": "Shunde Innovation School, University of Science and Technology Beijing, Foshan, 528399, China", "institution_ids": ["https://openalex.org/I92403157"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5023215301", "display_name": "Yuqi Yuan", "orcid": "https://orcid.org/0009-0001-9630-9074", "observed_orcids": ["https://orcid.org/0009-0001-9630-9074"]}, "institutions": [{"id": "https://openalex.org/I92403157", "display_name": "University of Science and Technology Beijing", "ror": "https://ror.org/02egmk993", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I92403157"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Yuqi Yuan", "raw_affiliation_strings": ["Beijing Key Laboratory of Knowledge Engineering for Materials Science, Beijing, 100083, China", "School of Computer and Communication Engineering, University of Science and Technology Beijing, Beijing, 100083, China", "Shunde Innovation School, University of Science and Technology Beijing, Foshan, 528399, China"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Beijing Key Laboratory of Knowledge Engineering for Materials Science, Beijing, 100083, China", "institution_ids": []}, {"raw_affiliation_string": "School of Computer and Communication Engineering, University of Science and Technology Beijing, Beijing, 100083, China", "institution_ids": ["https://openalex.org/I92403157"]}, {"raw_affiliation_string": "Shunde Innovation School, University of Science and Technology Beijing, Foshan, 528399, China", "institution_ids": ["https://openalex.org/I92403157"]}]}, {"author_position": "middle", "author": {"id": null, "display_name": "Fengbo Gu", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Fengbo Gu", "raw_affiliation_strings": ["Ouyeel Co., Ltd., Shanghai, 201999, China"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Ouyeel Co., Ltd., Shanghai, 201999, China", "institution_ids": []}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5088963097", "display_name": "Hailun Shen", "orcid": "https://orcid.org/0009-0003-9829-3250", "observed_orcids": ["https://orcid.org/0009-0003-9829-3250"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Hailun Shen", "raw_affiliation_strings": ["Ouyeel Co., Ltd., Shanghai, 201999, China"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Ouyeel Co., Ltd., Shanghai, 201999, China", "institution_ids": []}]}, {"author_position": "last", "author": {"id": "https://openalex.org/A5115595810", "display_name": "Ziyang Huang", "orcid": "https://orcid.org/0009-0000-8777-863X", "observed_orcids": ["https://orcid.org/0009-0000-8777-863X", "https://orcid.org/0009-0000-4570-8748"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Ziyang Huang", "raw_affiliation_strings": ["Ouyeel Co., Ltd., Shanghai, 201999, China"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Ouyeel Co., Ltd., Shanghai, 201999, China", "institution_ids": []}]}], "abstract_inverted_index": null, "primary_topic": {"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.8256999850273132, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.8256999850273132, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T11550", "display_name": "Text and Document Classification Technologies", "score": 0.07020000368356705, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10664", "display_name": "Sentiment Analysis and Opinion Mining", "score": 0.06610000133514404, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "doi:10.1007/s13042-025-02760-4", "is_oa": false, "landing_page_url": "https://doi.org/10.1007/s13042-025-02760-4", "pdf_url": null, "source": {"id": "https://openalex.org/S2764999920", "display_name": "International Journal of Machine Learning and Cybernetics", "issn_l": "1868-8071", "issn": ["1868-8071", "1868-808X"], "is_oa": false, "is_in_doaj": false, "is_core": true, "listed_in": ["cwts-core", "jufo-1", "norway-1"], "host_organization": "https://openalex.org/P4310319900", "host_organization_name": "Springer Science+Business Media", "host_organization_lineage": ["https://openalex.org/P4310319900", "https://openalex.org/P4310319965"], "host_organization_lineage_names": ["Springer Science+Business Media", "Springer Nature"], "type": "journal"}, "license": null, "license_id": null, "version": "publishedVersion", "is_accepted": true, "is_published": true, "raw_source_name": "International Journal of Machine Learning and Cybernetics", "raw_type": "journal-article"}, "cited_by_count": 3, "is_retracted": false}, {"id": "https://openalex.org/W4416223466", "doi": "https://doi.org/10.48550/arxiv.2511.09105", "title": "Cost-Minimized Label-Flipping Poisoning Attack to LLM Alignment", "publication_year": 2025, "publication_date": "2025-11-12", "type": "preprint", "language": "en", "authorships": [{"author_position": "first", "author": {"id": "https://openalex.org/A5120352487", "display_name": "Shigeki Kusaka", "orcid": null, "observed_orcids": []}, "institutions": [{"id": "https://openalex.org/I146399215", "display_name": "University of Tsukuba", "ror": "https://ror.org/02956yf07", "country_code": "JP", "type": "education", "lineage": ["https://openalex.org/I146399215"]}], "countries": ["JP"], "is_corresponding": false, "raw_author_name": "Kusaka, Shigeki", "raw_affiliation_strings": ["University of Tsukuba ,"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "University of Tsukuba ,", "institution_ids": ["https://openalex.org/I146399215"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5078368325", "display_name": "Keita Saito", "orcid": null, "observed_orcids": []}, "institutions": [{"id": "https://openalex.org/I146399215", "display_name": "University of Tsukuba", "ror": "https://ror.org/02956yf07", "country_code": "JP", "type": "education", "lineage": ["https://openalex.org/I146399215"]}], "countries": ["JP"], "is_corresponding": false, "raw_author_name": "Saito, Keita", "raw_affiliation_strings": ["University of Tsukuba ,"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "University of Tsukuba ,", "institution_ids": ["https://openalex.org/I146399215"]}]}, {"author_position": "middle", "author": {"id": null, "display_name": "Kudo, Mikoto", "orcid": null, "observed_orcids": []}, "institutions": [{"id": "https://openalex.org/I146399215", "display_name": "University of Tsukuba", "ror": "https://ror.org/02956yf07", "country_code": "JP", "type": "education", "lineage": ["https://openalex.org/I146399215"]}, {"id": "https://openalex.org/I4210110652", "display_name": "RIKEN", "ror": "https://ror.org/01sjwvz98", "country_code": "JP", "type": "facility", "lineage": ["https://openalex.org/I4210110652"]}, {"id": "https://openalex.org/I4210126580", "display_name": "RIKEN Center for Advanced Intelligence Project", "ror": "https://ror.org/03ckxwf91", "country_code": "JP", "type": "facility", "lineage": ["https://openalex.org/I4210110652", "https://openalex.org/I4210126580"]}], "countries": ["JP"], "is_corresponding": false, "raw_author_name": "Kudo, Mikoto", "raw_affiliation_strings": ["RIKEN AIP ,", "University of Tsukuba ,"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "RIKEN AIP ,", "institution_ids": ["https://openalex.org/I4210110652", "https://openalex.org/I4210126580"]}, {"raw_affiliation_string": "University of Tsukuba ,", "institution_ids": ["https://openalex.org/I146399215"]}]}, {"author_position": "middle", "author": {"id": null, "display_name": "Tanabe, Takumi", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Tanabe, Takumi", "raw_affiliation_strings": ["LY Corporation ,"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "LY Corporation ,", "institution_ids": []}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5052503437", "display_name": "Akifumi Wachi", "orcid": "https://orcid.org/0009-0008-4722-9861", "observed_orcids": ["https://orcid.org/0009-0008-4722-9861"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Wachi, Akifumi", "raw_affiliation_strings": ["LY Corporation ,"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "LY Corporation ,", "institution_ids": []}]}, {"author_position": "last", "author": {"id": "https://openalex.org/A5038757231", "display_name": "Youhei Akimoto", "orcid": "https://orcid.org/0000-0003-2760-8123", "observed_orcids": ["https://orcid.org/0000-0003-2760-8123"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Akimoto, Youhei", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}], "abstract_inverted_index": {"Large": [0], "language": [1], "models": [2], "(LLMs)": [3], "are": [4], "increasingly": [5], "deployed": [6], "in": [7, 149], "real-world": [8], "systems,": [9], "making": [10], "it": [11], "critical": [12], "to": [13, 40, 102, 140, 155], "understand": [14], "their": [15, 28, 157], "vulnerabilities.": [16], "While": [17], "data": [18], "poisoning": [19, 37, 114, 126, 161], "attacks": [20], "during": [21, 53], "RLHF/DPO": [22, 150], "alignment": [23], "have": [24], "been": [25], "studied": [26], "empirically,": [27], "theoretical": [29, 86], "foundations": [30], "remain": [31], "unclear.": [32], "We": [33, 60], "investigate": [34], "the": [35, 57, 77, 104, 112, 132, 141], "minimum-cost": [36], "attack": [38, 79, 94], "required": [39, 109], "steer": [41], "an": [42, 46], "LLM's": [43], "policy": [44], "toward": [45], "attacker's": [47], "target": [48], "by": [49], "flipping": [50], "preference": [51], "labels": [52], "RLHF/DPO,": [54], "without": [55], "altering": [56], "compared": [58], "outputs.": [59], "formulate": [61], "this": [62, 85, 120], "as": [63], "a": [64, 82], "convex": [65], "optimization": [66], "problem": [67], "with": [68], "linear": [69], "constraints,": [70], "deriving": [71], "lower": [72], "and": [73, 152], "upper": [74], "bounds": [75], "on": [76], "minimum": [78], "cost.": [80], "As": [81], "byproduct": [83], "of": [84, 106], "analysis,": [87], "we": [88], "show": [89], "that": [90, 119], "any": [91], "existing": [92], "label-flipping": [93], "can": [95, 123], "be": [96], "post-processed": [97], "via": [98], "our": [99], "proposed": [100], "method": [101], "reduce": [103, 125], "number": [105], "label": [107], "flips": [108], "while": [110], "preserving": [111], "intended": [113], "effect.": [115], "Empirical": [116], "results": [117], "demonstrate": [118], "cost-minimization": [121], "post-processing": [122], "significantly": [124], "costs": [127], "over": [128], "baselines,": [129], "particularly": [130], "when": [131], "reward": [133], "model's": [134], "feature": [135], "dimension": [136], "is": [137], "small": [138], "relative": [139], "dataset": [142], "size.": [143], "These": [144], "findings": [145], "highlight": [146], "fundamental": [147], "vulnerabilities": [148], "pipelines": [151], "provide": [153], "tools": [154], "evaluate": [156], "robustness": [158], "against": [159], "low-cost": [160], "attacks.": [162]}, "primary_topic": {"id": "https://openalex.org/T11689", "display_name": "Adversarial Robustness in Machine Learning", "score": 1.0, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T11689", "display_name": "Adversarial Robustness in Machine Learning", "score": 1.0, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10764", "display_name": "Privacy-Preserving Technologies in Data", "score": 0.0, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.0, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "pmh:oai:arXiv.org:2511.09105", "is_oa": true, "landing_page_url": "http://arxiv.org/abs/2511.09105", "pdf_url": "https://arxiv.org/pdf/2511.09105", "source": {"id": "https://openalex.org/S4306400194", "display_name": "arXiv (Cornell University)", "issn_l": "2331-8422", "issn": ["2331-8422"], "is_oa": true, "is_in_doaj": false, "is_core": false, "listed_in": [], "host_organization": "https://openalex.org/I205783295", "host_organization_name": "Cornell University", "host_organization_lineage": ["https://openalex.org/I205783295"], "host_organization_lineage_names": [], "type": "repository"}, "license": "cc-by", "license_id": "https://openalex.org/licenses/cc-by", "version": "submittedVersion", "is_accepted": false, "is_published": false, "raw_source_name": null, "raw_type": "text"}, "cited_by_count": 0, "is_retracted": false}, {"id": "https://openalex.org/W4414942664", "doi": "https://doi.org/10.48550/arxiv.2507.07817", "title": "On the Effect of Instruction Tuning Loss on Generalization", "publication_year": 2025, "publication_date": "2025-07-10", "type": "preprint", "language": "en", "authorships": [{"author_position": "first", "author": {"id": "https://openalex.org/A5104707682", "display_name": "Anwoy Chatterjee", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Chatterjee, Anwoy", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5091932316", "display_name": "H S V N S Kowndinya Renduchintala", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Renduchintala, H S V N S Kowndinya", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5022961965", "display_name": "Sumit Bhatia", "orcid": "https://orcid.org/0000-0002-8146-4100", "observed_orcids": ["https://orcid.org/0000-0002-8146-4100"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Bhatia, Sumit", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "last", "author": {"id": "https://openalex.org/A5046521217", "display_name": "Tanmoy Chakraborty", "orcid": "https://orcid.org/0000-0002-0210-0369", "observed_orcids": ["https://orcid.org/0000-0002-0210-0369"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Chakraborty, Tanmoy", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}], "abstract_inverted_index": {"Instruction": [0, 86], "Tuning": [1, 87], "has": [2, 24], "emerged": [3], "as": [4, 89, 165], "a": [5, 90, 142, 150], "pivotal": [6], "post-training": [7], "paradigm": [8], "that": [9, 122, 141], "enables": [10], "pre-trained": [11], "language": [12, 102], "models": [13, 103, 159], "to": [14, 27, 93, 135, 180], "better": [15, 91, 166], "follow": [16], "user": [17], "instructions.": [18], "Despite": [19], "its": [20], "significance,": [21], "little": [22], "attention": [23], "been": [25], "given": [26], "optimizing": [28], "the": [29, 41, 70, 123, 157, 170, 178], "loss": [30, 47, 127, 184], "function": [31], "used.": [32], "A": [33], "fundamental,": [34], "yet": [35], "often": [36, 128], "overlooked,": [37], "question": [38], "is": [39, 48, 58, 198], "whether": [40], "conventional": [42, 94], "auto-regressive": [43], "objective": [44], "-": [45, 57], "where": [46], "computed": [49], "only": [50], "on": [51, 100], "response": [52, 77, 154], "tokens,": [53], "excluding": [54], "prompt": [55, 75, 137, 146], "tokens": [56, 78, 147, 155], "truly": [59], "optimal": [60], "for": [61, 145, 153, 169, 189], "instruction": [62, 80, 95, 125, 182], "tuning.": [63, 96], "In": [64], "this": [65], "work,": [66], "we": [67, 120], "systematically": [68], "investigate": [69], "impact": [71], "of": [72, 104, 112], "differentially": [73], "weighting": [74], "and": [76, 83, 107, 115, 132, 162, 185, 193], "in": [79], "tuning": [81, 126, 183], "loss,": [82], "propose": [84], "Weighted": [85], "(WIT)": [88], "alternative": [92], "Through": [97], "extensive": [98], "experiments": [99], "five": [101, 116], "different": [105, 113], "families": [106], "scale,": [108], "three": [109], "finetuning": [110], "datasets": [111], "sizes,": [114], "diverse": [117], "evaluation": [118], "benchmarks,": [119], "show": [121], "standard": [124], "yields": [129, 156], "suboptimal": [130], "performance": [131], "limited": [133], "robustness": [134], "input": [136], "variations.": [138], "We": [139], "find": [140], "low-to-moderate": [143], "weight": [144, 152], "coupled": [148], "with": [149], "moderate-to-high": [151], "best-performing": [158], "across": [160], "settings": [161], "also": [163], "serve": [164], "starting": [167], "points": [168], "subsequent": [171], "preference": [172], "alignment": [173], "training.": [174], "These": [175], "findings": [176], "highlight": [177], "need": [179], "reconsider": [181], "offer": [186], "actionable": [187], "insights": [188], "developing": [190], "more": [191], "robust": [192], "generalizable": [194], "models.": [195], "Our": [196], "code": [197], "open-sourced": [199], "at": [200], "https://github.com/kowndinya-renduchintala/WIT.": [201]}, "primary_topic": {"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.8217999935150146, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.8217999935150146, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10181", "display_name": "Natural Language Processing Techniques", "score": 0.07639999687671661, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T11307", "display_name": "Domain Adaptation and Few-Shot Learning", "score": 0.06790000200271606, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "pmh:oai:arXiv.org:2507.07817", "is_oa": true, "landing_page_url": "http://arxiv.org/abs/2507.07817", "pdf_url": "https://arxiv.org/pdf/2507.07817", "source": {"id": "https://openalex.org/S4306400194", "display_name": "arXiv (Cornell University)", "issn_l": "2331-8422", "issn": ["2331-8422"], "is_oa": true, "is_in_doaj": false, "is_core": false, "listed_in": [], "host_organization": "https://openalex.org/I205783295", "host_organization_name": "Cornell University", "host_organization_lineage": ["https://openalex.org/I205783295"], "host_organization_lineage_names": [], "type": "repository"}, "license": null, "license_id": null, "version": "submittedVersion", "is_accepted": false, "is_published": false, "raw_source_name": null, "raw_type": "text"}, "cited_by_count": 0, "is_retracted": false}, {"id": "https://openalex.org/W4410087239", "doi": "https://doi.org/10.1109/access.2025.3567244", "title": "ResilioMate: A Resilient Multi-Agent Task Executing Framework for Enhancing Small Language Models", "publication_year": 2025, "publication_date": "2025-01-01", "type": "article", "language": "en", "authorships": [{"author_position": "first", "author": {"id": "https://openalex.org/A5032489133", "display_name": "Yubing Xiong", "orcid": null, "observed_orcids": []}, "institutions": [{"id": "https://openalex.org/I13985625", "display_name": "East China Jiaotong University", "ror": "https://ror.org/05x2f1m38", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I13985625"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Yubing Xiong", "raw_affiliation_strings": ["School of Transportation Engineering, East China Jiaotong University, Nanchang, Jiangxi, China"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "School of Transportation Engineering, East China Jiaotong University, Nanchang, Jiangxi, China", "institution_ids": ["https://openalex.org/I13985625"]}]}, {"author_position": "middle", "author": {"id": null, "display_name": "Mingrui Huang", "orcid": "https://orcid.org/0009-0009-9933-2315", "observed_orcids": ["https://orcid.org/0009-0009-9933-2315"]}, "institutions": [{"id": "https://openalex.org/I13985625", "display_name": "East China Jiaotong University", "ror": "https://ror.org/05x2f1m38", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I13985625"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Mingrui Huang", "raw_affiliation_strings": ["School of Transportation Engineering, East China Jiaotong University, Nanchang, Jiangxi, China"], "raw_orcid": "https://orcid.org/0009-0009-9933-2315", "affiliations": [{"raw_affiliation_string": "School of Transportation Engineering, East China Jiaotong University, Nanchang, Jiangxi, China", "institution_ids": ["https://openalex.org/I13985625"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5101294363", "display_name": "Xuechen Liang", "orcid": "https://orcid.org/0009-0005-2318-6746", "observed_orcids": ["https://orcid.org/0009-0005-2318-6746"]}, "institutions": [{"id": "https://openalex.org/I13985625", "display_name": "East China Jiaotong University", "ror": "https://ror.org/05x2f1m38", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I13985625"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Xuechen Liang", "raw_affiliation_strings": ["School of Transportation Engineering, East China Jiaotong University, Nanchang, Jiangxi, China"], "raw_orcid": "https://orcid.org/0009-0005-2318-6746", "affiliations": [{"raw_affiliation_string": "School of Transportation Engineering, East China Jiaotong University, Nanchang, Jiangxi, China", "institution_ids": ["https://openalex.org/I13985625"]}]}, {"author_position": "last", "author": {"id": "https://openalex.org/A5093751229", "display_name": "Meiling Tao", "orcid": "https://orcid.org/0009-0003-8403-395X", "observed_orcids": ["https://orcid.org/0009-0003-8403-395X"]}, "institutions": [{"id": "https://openalex.org/I139024713", "display_name": "Guangdong University of Technology", "ror": "https://ror.org/04azbjn80", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I139024713"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Meiling Tao", "raw_affiliation_strings": ["School of Computer Science and Technology, Guangdong University of Technology, Guangzhou, China"], "raw_orcid": "https://orcid.org/0009-0003-8403-395X", "affiliations": [{"raw_affiliation_string": "School of Computer Science and Technology, Guangdong University of Technology, Guangzhou, China", "institution_ids": ["https://openalex.org/I139024713"]}]}], "abstract_inverted_index": {"Recent": [0], "advances": [1], "in": [2, 28, 126, 144, 195, 215], "large": [3], "language": [4, 21, 213], "models": [5, 22], "(LLMs)": [6], "have": [7], "been": [8], "limited": [9], "by": [10, 42, 180], "their": [11], "processing": [12], "requirements": [13], "and": [14, 51, 72, 99, 186, 202, 217], "vulnerability": [15], "to": [16, 68, 103, 148], "adversarial": [17, 162], "assaults,": [18], "whilst": [19], "short": [20], "(SLMs)": [23], "struggle": [24], "with": [25, 77, 133], "performance": [26, 125, 168, 198], "consistency": [27], "complex": [29], "tasks.": [30], "This": [31], "research": [32], "introduces": [33], "ResilioMate,": [34], "a": [35, 140, 205], "resilient": [36], "multi-agent": [37, 187], "framework": [38], "that": [39], "enhances": [40], "SLMs": [41, 201], "utilizing": [43], "distributed": [44], "cognitive": [45, 153], "burden": [46], "distribution,": [47], "dual-scale": [48, 174], "memory": [49, 175], "systems,": [50], "collaborative": [52, 152], "bias": [53], "prevention": [54], "strategies.": [55], "The": [56, 117, 172], "method": [57], "employs": [58], "dynamic": [59, 100], "task": [60, 184], "decomposition": [61], "across": [62], "specialized": [63], "agents": [64, 214], "(e.g.,": [65], "Assistant,": [66], "Checker)": [67], "minimize": [69], "computational": [70, 178], "costs": [71], "combines": [73], "short-term": [74], "trajectory": [75], "tracking": [76, 185], "long-term": [78], "self-reflective": [79], "optimization": [80], "for": [81, 208], "adaptive": [82], "execution.": [83], "At": [84], "its": [85], "core,": [86], "the": [87, 197, 209], "LeptoConnect": [88, 119], "model": [89, 120], "series": [90], "(1.8B/7B": [91], "parameters)": [92], "is": [93], "created": [94], "using": [95], "hybrid": [96], "attention": [97], "distillation": [98], "curriculum": [101], "learning": [102], "enable": [104], "cross-domain": [105], "competence": [106], "while": [107, 164], "retaining": [108], "SLM": [109], "efficiency.": [110], "ResilioMate": [111], "accomplishes": [112], "three": [113], "critical": [114], "improvements:": [115], "(1)": [116], "1.8B": [118], "attains": [121], "81.6%": [122], "of": [123, 142, 169, 211], "GPT-4\u2019s": [124, 149], "knowledge": [127, 188], "graph": [128], "construction": [129], "through": [130, 151, 182], "parameter-efficient": [131], "fine-tuning": [132], "structured": [134], "weight": [135], "matrices;": [136], "(2)": [137], "LeptoConnect-7B": [138], "achieves": [139], "score": [141], "41.3": [143], "database": [145], "operations,": [146], "compared": [147], "32.0,": [150], "load": [154], "allocation;": [155], "(3)": [156], "A": [157], "bias-interception": [158], "network": [159], "effectively": [160], "suppresses": [161], "propagation": [163], "achieving": [165], "code": [166], "correction": [167], "ROUGE-L\u2019s": [170], "42.86.": [171], "framework\u2019s": [173], "architecture": [176], "reduces": [177], "redundancy": [179], "26.4%": [181], "real-time": [183, 216], "refinement.": [189], "These": [190], "developments": [191], "illustrate": [192], "ResilioMate\u2019s": [193], "efficacy": [194], "bridging": [196], "gap": [199], "between": [200], "LLMs,": [203], "offering": [204], "scalable": [206], "solution": [207], "deployment": [210], "efficient": [212], "peripheral": [218], "computing": [219], "environments.": [220]}, "primary_topic": {"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.7202000021934509, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.7202000021934509, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10181", "display_name": "Natural Language Processing Techniques", "score": 0.18629999458789825, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T11307", "display_name": "Domain Adaptation and Few-Shot Learning", "score": 0.013500000350177288, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "doi:10.1109/access.2025.3567244", "is_oa": true, "landing_page_url": "https://doi.org/10.1109/access.2025.3567244", "pdf_url": null, "source": {"id": "https://openalex.org/S2485537415", "display_name": "IEEE Access", "issn_l": "2169-3536", "issn": ["2169-3536"], "is_oa": true, "is_in_doaj": true, "is_core": true, "listed_in": ["cwts-core", "doaj", "jufo-1", "ki-jl-1", "norway-1"], "host_organization": "https://openalex.org/P4310319808", "host_organization_name": "Institute of Electrical and Electronics Engineers", "host_organization_lineage": ["https://openalex.org/P4310319808"], "host_organization_lineage_names": ["Institute of Electrical and Electronics Engineers"], "type": "journal"}, "license": "cc-by", "license_id": "https://openalex.org/licenses/cc-by", "version": "publishedVersion", "is_accepted": true, "is_published": true, "raw_source_name": "IEEE Access", "raw_type": "journal-article"}, "cited_by_count": 4, "is_retracted": false}, {"id": "https://openalex.org/W4410830113", "doi": "https://doi.org/10.1145/3736759", "title": "Open-ended Autoregressive Visual Storytelling via Parameter Efficient Instruction Tuning", "publication_year": 2025, "publication_date": "2025-05-28", "type": "article", "language": "en", "authorships": [{"author_position": "first", "author": {"id": null, "display_name": "Lixin Liu", "orcid": "https://orcid.org/0009-0000-2281-9939", "observed_orcids": ["https://orcid.org/0009-0000-2281-9939"]}, "institutions": [{"id": "https://openalex.org/I53592917", "display_name": "Jiangxi Normal University", "ror": "https://ror.org/05nkgk822", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I53592917"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Lixin Liu", "raw_affiliation_strings": ["School of Digital Industry, Jiangxi Normal University", "School of Digital Industry, Jiangxi Normal University, Shangrao, China"], "raw_orcid": "https://orcid.org/0009-0000-2281-9939", "affiliations": [{"raw_affiliation_string": "School of Digital Industry, Jiangxi Normal University", "institution_ids": ["https://openalex.org/I53592917"]}, {"raw_affiliation_string": "School of Digital Industry, Jiangxi Normal University, Shangrao, China", "institution_ids": ["https://openalex.org/I53592917"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5034126758", "display_name": "Aiwen Jiang", "orcid": "https://orcid.org/0000-0002-5979-7590", "observed_orcids": ["https://orcid.org/0000-0002-5979-7590"]}, "institutions": [{"id": "https://openalex.org/I53592917", "display_name": "Jiangxi Normal University", "ror": "https://ror.org/05nkgk822", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I53592917"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Aiwen Jiang", "raw_affiliation_strings": ["School of Digital Industry, Jiangxi Normal University", "School of Digital Industry, Jiangxi Normal University, Shangrao, China"], "raw_orcid": "https://orcid.org/0000-0002-5979-7590", "affiliations": [{"raw_affiliation_string": "School of Digital Industry, Jiangxi Normal University", "institution_ids": ["https://openalex.org/I53592917"]}, {"raw_affiliation_string": "School of Digital Industry, Jiangxi Normal University, Shangrao, China", "institution_ids": ["https://openalex.org/I53592917"]}]}, {"author_position": "middle", "author": {"id": null, "display_name": "Yinuo Chen", "orcid": "https://orcid.org/0009-0007-9217-1817", "observed_orcids": ["https://orcid.org/0009-0007-9217-1817"]}, "institutions": [{"id": "https://openalex.org/I53592917", "display_name": "Jiangxi Normal University", "ror": "https://ror.org/05nkgk822", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I53592917"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Yinuo Chen", "raw_affiliation_strings": ["School of Digital Industry, Jiangxi Normal University", "School of Digital Industry, Jiangxi Normal University, Shangrao, China"], "raw_orcid": "https://orcid.org/0009-0007-9217-1817", "affiliations": [{"raw_affiliation_string": "School of Digital Industry, Jiangxi Normal University", "institution_ids": ["https://openalex.org/I53592917"]}, {"raw_affiliation_string": "School of Digital Industry, Jiangxi Normal University, Shangrao, China", "institution_ids": ["https://openalex.org/I53592917"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5101859009", "display_name": "Changhong Liu", "orcid": "https://orcid.org/0000-0003-4673-5806", "observed_orcids": ["https://orcid.org/0000-0003-4673-5806"]}, "institutions": [{"id": "https://openalex.org/I53592917", "display_name": "Jiangxi Normal University", "ror": "https://ror.org/05nkgk822", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I53592917"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Changhong Liu", "raw_affiliation_strings": ["School of Computer and Infomation Engineering, Jiangxi Normal University", "School of Computer and Infomation Engineering, Jiangxi Normal University, Nanchang, China"], "raw_orcid": "https://orcid.org/0000-0003-4673-5806", "affiliations": [{"raw_affiliation_string": "School of Computer and Infomation Engineering, Jiangxi Normal University", "institution_ids": ["https://openalex.org/I53592917"]}, {"raw_affiliation_string": "School of Computer and Infomation Engineering, Jiangxi Normal University, Nanchang, China", "institution_ids": ["https://openalex.org/I53592917"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5112378738", "display_name": "Qi Huang", "orcid": "https://orcid.org/0000-0002-3603-1875", "observed_orcids": ["https://orcid.org/0000-0002-3603-1875"]}, "institutions": [{"id": "https://openalex.org/I53592917", "display_name": "Jiangxi Normal University", "ror": "https://ror.org/05nkgk822", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I53592917"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Qi Huang", "raw_affiliation_strings": ["School of Computer and Information Engineering, Jiangxi Normal University", "School of Computer and Information Engineering, Jiangxi Normal University, Nanchang, China"], "raw_orcid": "https://orcid.org/0000-0002-3603-1875", "affiliations": [{"raw_affiliation_string": "School of Computer and Information Engineering, Jiangxi Normal University", "institution_ids": ["https://openalex.org/I53592917"]}, {"raw_affiliation_string": "School of Computer and Information Engineering, Jiangxi Normal University, Nanchang, China", "institution_ids": ["https://openalex.org/I53592917"]}]}, {"author_position": "last", "author": {"id": "https://openalex.org/A5033849258", "display_name": "Mingwen Wang", "orcid": "https://orcid.org/0000-0002-8118-9612", "observed_orcids": ["https://orcid.org/0000-0002-8118-9612"]}, "institutions": [{"id": "https://openalex.org/I53592917", "display_name": "Jiangxi Normal University", "ror": "https://ror.org/05nkgk822", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I53592917"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Mingwen Wang", "raw_affiliation_strings": ["School of Digital Industry, Jiangxi Normal University", "School of Digital Industry, Jiangxi Normal University, Shangrao, China"], "raw_orcid": "https://orcid.org/0000-0002-8118-9612", "affiliations": [{"raw_affiliation_string": "School of Digital Industry, Jiangxi Normal University", "institution_ids": ["https://openalex.org/I53592917"]}, {"raw_affiliation_string": "School of Digital Industry, Jiangxi Normal University, Shangrao, China", "institution_ids": ["https://openalex.org/I53592917"]}]}], "abstract_inverted_index": {"Visual": [0], "storytelling": [1, 25], "(VIST)": [2], "involves": [3], "generating": [4], "coherent,": [5], "creative,": [6], "and": [7, 73, 87, 99, 104, 174, 182, 201], "vivid": [8, 88, 125], "narrative": [9], "for": [10, 127, 139], "a": [11, 118], "collection": [12], "of": [13, 162, 178], "images.": [14], "It": [15, 155], "remains": [16], "an": [17, 147, 166], "immense": [18], "challenge": [19], "within": [20], "cross-modal": [21, 45], "domain.": [22], "Traditional": [23], "mainstream": [24], "work": [26], "were": [27, 51], "less": [28], "proficient": [29], "in": [30, 65, 91, 165, 176], "handling": [31, 56], "long": [32], "sequential": [33], "relationships.": [34], "Though": [35], "large-scale": [36], "visual-language": [37], "pre-training": [38], "(VLP)": [39], "models": [40, 81, 202], "demonstrated": [41, 189], "promising": [42], "prospect": [43], "on": [44, 135, 205], "tasks.": [46], "So": [47], "far": [48], "they": [49], "still": [50], "not": [52], "particularly": [53], "adept": [54], "at": [55], "tasks": [57], "involving": [58], "image": [59, 129, 160], "sequences.": [60, 130], "Moreover,": [61], "the": [62, 66], "reference": [63], "descriptions": [64], "available": [67], "VIST": [68], "benchmark": [69], "dataset": [70, 103, 120, 132], "are": [71, 203], "short": [72], "simplistic,": [74], "which": [75, 122], "constrains": [76], "model\u2019s": [77], "potential": [78, 134], "capabilities.": [79], "Current": [80], "struggle": [82], "to": [83, 159], "produce": [84], "truly": [85], "rich": [86], "narratives.": [89], "Therefore,": [90], "this": [92], "article,": [93], "we": [94, 115, 144], "will": [95], "address": [96], "these": [97], "deficiencies,": [98], "contribute": [100], "from": [101], "both": [102], "innovative": [105, 148], "model": [106, 113, 141, 152], "aspects.": [107], "Firstly,": [108], "by": [109], "leveraging": [110], "large": [111], "language": [112], "(LLM),": [114], "have": [116, 145, 170, 187], "constructed": [117], "new": [119], "VIST++": [121], "can": [123, 156], "enrich": [124], "narratives": [126], "open-ended": [128, 167], "The": [131, 184], "has": [133], "providing": [136], "beneficial": [137], "support": [138], "future": [140], "learning.": [142], "Secondly,": [143], "proposed": [146], "auto-regressive": [149], "story": [150], "generation": [151], "named": [153], "ReStoryGen.": [154], "be": [157], "applied": [158], "sequences": [161], "varying": [163], "lengths": [164], "way.": [168], "We": [169], "performed": [171], "extensive": [172], "experiments": [173], "evaluations": [175], "terms": [177], "visual": [179], "grounding,": [180], "coherence": [181], "non-redundancy.": [183], "experiment": [185], "results": [186], "convincingly": [188], "ReStoryGen": [190], "achieves": [191], "impressive": [192], "outcomes": [193], "through": [194], "utilizing": [195], "parameter-efficient": [196], "instruction-tuning.": [197], "Related": [198], "source": [199], "codes": [200], "distributed": [204], "Github": [206], "https://github.com/lixinliu1995/story_gen": [207], ".": [208]}, "primary_topic": {"id": "https://openalex.org/T11714", "display_name": "Multimodal Machine Learning Applications", "score": 1.0, "subfield": {"id": "https://openalex.org/subfields/1707", "display_name": "Computer Vision and Pattern Recognition"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T11714", "display_name": "Multimodal Machine Learning Applications", "score": 1.0, "subfield": {"id": "https://openalex.org/subfields/1707", "display_name": "Computer Vision and Pattern Recognition"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10775", "display_name": "Generative Adversarial Networks and Image Synthesis", "score": 0.0, "subfield": {"id": "https://openalex.org/subfields/1707", "display_name": "Computer Vision and Pattern Recognition"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.0, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "doi:10.1145/3736759", "is_oa": true, "landing_page_url": "https://doi.org/10.1145/3736759", "pdf_url": "https://dl.acm.org/doi/pdf/10.1145/3736759", "source": {"id": "https://openalex.org/S4306421405", "display_name": "ACM Transactions on Asian and Low-Resource Language Information Processing", "issn_l": "2375-4699", "issn": ["2375-4699", "2375-4702"], "is_oa": false, "is_in_doaj": false, "is_core": true, "listed_in": ["cwts-core", "jufo-1", "norway-1"], "host_organization": "https://openalex.org/P4310319798", "host_organization_name": "Association for Computing Machinery", "host_organization_lineage": ["https://openalex.org/P4310319798"], "host_organization_lineage_names": ["Association for Computing Machinery"], "type": "journal"}, "license": null, "license_id": null, "version": "publishedVersion", "is_accepted": true, "is_published": true, "raw_source_name": "ACM Transactions on Asian and Low-Resource Language Information Processing", "raw_type": "journal-article"}, "cited_by_count": 0, "is_retracted": false}, {"id": "https://openalex.org/W4410351910", "doi": "https://doi.org/10.1177/1088467x251340666", "title": "Research on hospital information management system based on large language model (LLM) algorithm model", "publication_year": 2025, "publication_date": "2025-05-14", "type": "article", "language": "en", "authorships": [{"author_position": "first", "author": {"id": "https://openalex.org/A5101629976", "display_name": "Fan Zhang", "orcid": "https://orcid.org/0000-0003-0120-4892", "observed_orcids": ["https://orcid.org/0000-0003-0120-4892"]}, "institutions": [{"id": "https://openalex.org/I17721919", "display_name": "Shanxi Medical University", "ror": "https://ror.org/0265d1010", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I17721919"]}, {"id": "https://openalex.org/I4210125748", "display_name": "First Hospital of Shanxi Medical University", "ror": "https://ror.org/02vzqaq35", "country_code": "CN", "type": "healthcare", "lineage": ["https://openalex.org/I4210125748"]}], "countries": ["CN"], "is_corresponding": true, "raw_author_name": "Zhang Fan", "raw_affiliation_strings": ["The First Hospital of Shanxi Medical University, Taiyuan City, Shanxi Province, 030001, China"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "The First Hospital of Shanxi Medical University, Taiyuan City, Shanxi Province, 030001, China", "institution_ids": ["https://openalex.org/I17721919", "https://openalex.org/I4210125748"]}]}, {"author_position": "last", "author": {"id": null, "display_name": "Wang Jiyu", "orcid": null, "observed_orcids": []}, "institutions": [{"id": "https://openalex.org/I17721919", "display_name": "Shanxi Medical University", "ror": "https://ror.org/0265d1010", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I17721919"]}, {"id": "https://openalex.org/I4210125748", "display_name": "First Hospital of Shanxi Medical University", "ror": "https://ror.org/02vzqaq35", "country_code": "CN", "type": "healthcare", "lineage": ["https://openalex.org/I4210125748"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Wang Jiyu", "raw_affiliation_strings": ["The First Hospital of Shanxi Medical University, Taiyuan City, Shanxi Province, 030001, China"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "The First Hospital of Shanxi Medical University, Taiyuan City, Shanxi Province, 030001, China", "institution_ids": ["https://openalex.org/I17721919", "https://openalex.org/I4210125748"]}]}], "abstract_inverted_index": {"In": [0], "recent": [1], "years,": [2], "the": [3, 43, 47, 52, 96, 146, 166, 175], "application": [4], "of": [5, 46, 59, 122, 128, 169], "artificial": [6], "intelligence,": [7], "particularly": [8], "generative": [9, 33, 150, 170], "AI": [10, 171], "(AIGC),": [11], "has": [12], "gained": [13], "significant": [14], "traction": [15], "across": [16], "various": [17], "industries.": [18], "This": [19, 68, 163], "study": [20, 164], "focuses": [21], "on": [22], "employing": [23], "a": [24, 32, 80], "Large": [25], "Language": [26], "Model": [27, 85], "(LLM)": [28], "algorithm": [29], "to": [30, 78], "construct": [31], "dialogue": [34, 82], "model": [35, 147], "specifically": [36], "tailored": [37], "for": [38, 177], "traditional": [39, 153], "Chinese": [40, 48, 154], "medicine,": [41, 155], "utilizing": [42], "extensive": [44], "database": [45], "Medical": [49], "Code": [50], "as": [51, 118], "primary": [53], "data": [54, 72], "source.": [55], "The": [56, 109, 142], "methodology": [57], "consists": [58], "three": [60], "key": [61], "phases:": [62], "(1)": [63], "Corpus": [64], "Collection": [65], "and": [66, 74, 103, 106, 124, 140, 181], "Preprocessing:": [67], "involves": [69], "web": [70], "crawling,": [71], "cleaning,": [73], "segmenting": [75], "ancient": [76], "texts": [77], "prepare": [79], "comprehensive": [81], "dataset;": [83], "(2)": [84], "Training:": [86], "A": [87], "structured": [88], "approach": [89], "is": [90, 112], "applied": [91], "through": [92, 114], "continuous": [93], "pretraining": [94], "using": [95], "Ziya-LLaMA-13B-v1": [97], "base,": [98], "followed": [99], "by": [100], "supervised": [101], "fine-tuning": [102], "DPO": [104], "optimization;": [105], "(3)": [107], "Evaluation:": [108], "model\u2019s": [110], "performance": [111], "assessed": [113], "automatic": [115], "metrics": [116], "such": [117], "BLEU": [119], "(average": [120, 126], "score": [121, 127], "0.465)": [123], "ROUGE": [125], "0.392),": [129], "alongside": [130], "manual": [131], "evaluations": [132], "indicating": [133], "high": [134], "response": [135], "accuracy": [136], "(scores": [137], "between": [138], "8": [139], "9).": [141], "results": [143], "suggest": [144], "that": [145], "demonstrates": [148], "robust": [149], "capabilities": [151], "in": [152, 159, 172], "outperforming": [156], "existing": [157], "models": [158], "answering": [160], "knowledge": [161], "questions.": [162], "highlights": [165], "transformative": [167], "potential": [168], "healthcare,": [173], "paving": [174], "way": [176], "enhanced": [178], "hospital": [179], "management": [180], "patient": [182], "engagement": [183], "tools.": [184]}, "primary_topic": {"id": "https://openalex.org/T11636", "display_name": "Artificial Intelligence in Healthcare and Education", "score": 0.82669997215271, "subfield": {"id": "https://openalex.org/subfields/2718", "display_name": "Health Informatics"}, "field": {"id": "https://openalex.org/fields/27", "display_name": "Medicine"}, "domain": {"id": "https://openalex.org/domains/4", "display_name": "Health Sciences"}}, "topics": [{"id": "https://openalex.org/T11636", "display_name": "Artificial Intelligence in Healthcare and Education", "score": 0.82669997215271, "subfield": {"id": "https://openalex.org/subfields/2718", "display_name": "Health Informatics"}, "field": {"id": "https://openalex.org/fields/27", "display_name": "Medicine"}, "domain": {"id": "https://openalex.org/domains/4", "display_name": "Health Sciences"}}, {"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.11500000208616257, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T12031", "display_name": "Speech and dialogue systems", "score": 0.020600000396370888, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "doi:10.1177/1088467x251340666", "is_oa": false, "landing_page_url": "https://doi.org/10.1177/1088467x251340666", "pdf_url": null, "source": {"id": "https://openalex.org/S2498839158", "display_name": "Intelligent Data Analysis", "issn_l": "1088-467X", "issn": ["1088-467X", "1571-4128"], "is_oa": false, "is_in_doaj": false, "is_core": true, "listed_in": ["cwts-core", "jufo-1", "ki-jl-1", "norway-1"], "host_organization": "https://openalex.org/P4310318577", "host_organization_name": "IOS Press", "host_organization_lineage": ["https://openalex.org/P4310318577"], "host_organization_lineage_names": ["IOS Press"], "type": "journal"}, "license": null, "license_id": null, "version": "publishedVersion", "is_accepted": true, "is_published": true, "raw_source_name": "Intelligent Data Analysis: An International Journal", "raw_type": "journal-article"}, "cited_by_count": 0, "is_retracted": false}, {"id": "https://openalex.org/W4414633044", "doi": "https://doi.org/10.48550/arxiv.2504.15777", "title": "Tina: Tiny Reasoning Models via LoRA", "publication_year": 2025, "publication_date": "2025-04-22", "type": "preprint", "language": "en", "authorships": [{"author_position": "first", "author": {"id": "https://openalex.org/A5102785891", "display_name": "Shangshang Wang", "orcid": "https://orcid.org/0009-0005-0996-5380", "observed_orcids": ["https://orcid.org/0009-0005-0996-5380"]}, "institutions": [{"id": "https://openalex.org/I1174212", "display_name": "University of Southern California", "ror": "https://ror.org/03taz7m60", "country_code": "US", "type": "education", "lineage": ["https://openalex.org/I1174212"]}], "countries": ["US"], "is_corresponding": false, "raw_author_name": "Wang, Shangshang", "raw_affiliation_strings": ["University of Southern California"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "University of Southern California", "institution_ids": ["https://openalex.org/I1174212"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5092950123", "display_name": "Julian Asilis", "orcid": "https://orcid.org/0000-0002-7047-3861", "observed_orcids": ["https://orcid.org/0000-0002-7047-3861"]}, "institutions": [{"id": "https://openalex.org/I1174212", "display_name": "University of Southern California", "ror": "https://ror.org/03taz7m60", "country_code": "US", "type": "education", "lineage": ["https://openalex.org/I1174212"]}], "countries": ["US"], "is_corresponding": false, "raw_author_name": "Asilis, Julian", "raw_affiliation_strings": ["University of Southern California"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "University of Southern California", "institution_ids": ["https://openalex.org/I1174212"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5028059095", "display_name": "\u00d6mer Faruk Akg\u00fcl", "orcid": "https://orcid.org/0000-0003-1588-3670", "observed_orcids": ["https://orcid.org/0000-0003-1588-3670"]}, "institutions": [{"id": "https://openalex.org/I1174212", "display_name": "University of Southern California", "ror": "https://ror.org/03taz7m60", "country_code": "US", "type": "education", "lineage": ["https://openalex.org/I1174212"]}], "countries": ["US"], "is_corresponding": false, "raw_author_name": "Akg\u00fcl, \u00d6mer Faruk", "raw_affiliation_strings": ["University of Southern California"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "University of Southern California", "institution_ids": ["https://openalex.org/I1174212"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5109467016", "display_name": "Emrullah Bilgin", "orcid": null, "observed_orcids": []}, "institutions": [{"id": "https://openalex.org/I1174212", "display_name": "University of Southern California", "ror": "https://ror.org/03taz7m60", "country_code": "US", "type": "education", "lineage": ["https://openalex.org/I1174212"]}], "countries": ["US"], "is_corresponding": false, "raw_author_name": "Bilgin, Enes Burak", "raw_affiliation_strings": ["University of Southern California"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "University of Southern California", "institution_ids": ["https://openalex.org/I1174212"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5041927922", "display_name": "Ollie Liu", "orcid": null, "observed_orcids": []}, "institutions": [{"id": "https://openalex.org/I1174212", "display_name": "University of Southern California", "ror": "https://ror.org/03taz7m60", "country_code": "US", "type": "education", "lineage": ["https://openalex.org/I1174212"]}], "countries": ["US"], "is_corresponding": false, "raw_author_name": "Liu, Ollie", "raw_affiliation_strings": ["University of Southern California"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "University of Southern California", "institution_ids": ["https://openalex.org/I1174212"]}]}, {"author_position": "last", "author": {"id": "https://openalex.org/A5073310637", "display_name": "Willie Neiswanger", "orcid": "https://orcid.org/0000-0002-9619-5572", "observed_orcids": ["https://orcid.org/0000-0002-9619-5572"]}, "institutions": [{"id": "https://openalex.org/I1174212", "display_name": "University of Southern California", "ror": "https://ror.org/03taz7m60", "country_code": "US", "type": "education", "lineage": ["https://openalex.org/I1174212"]}], "countries": ["US"], "is_corresponding": false, "raw_author_name": "Neiswanger, Willie", "raw_affiliation_strings": ["University of Southern California"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "University of Southern California", "institution_ids": ["https://openalex.org/I1174212"]}]}], "abstract_inverted_index": {"How": [0], "cost-effectively": [1], "can": [2, 36], "strong": [3], "reasoning": [4, 23, 34, 70, 81, 116, 148, 157, 191], "abilities": [5], "be": [6, 37], "achieved": [7, 25, 92], "in": [8], "language": [9], "models?": [10], "Driven": [11], "by": [12, 43, 103, 193], "this": [13, 90, 153, 175], "fundamental": [14], "question,": [15], "we": [16, 172, 210], "present": [17], "Tina,": [18], "a": [19, 94, 114, 165], "family": [20], "of": [21, 97, 145, 169, 190, 205], "tiny": [22, 58, 95], "models": [24, 67, 82], "with": [26, 164], "high": [27], "cost-efficiency.": [28], "Notably,": [29], "Tina": [30, 111], "demonstrates": [31], "that": [32, 68, 174], "substantial": [33], "performance": [35, 71, 117], "developed": [38], "using": [39, 51], "only": [40, 126], "minimal": [41], "resources,": [42], "applying": [44], "parameter-efficient": [45], "updates": [46], "during": [47], "reinforcement": [48], "learning": [49], "(RL),": [50], "low-rank": [52], "adaptation": [53], "(LoRA),": [54], "to": [55, 186], "an": [56, 134], "already": [57], "1.5B": [59], "parameter": [60], "base": [61, 87, 199], "model.": [62, 88], "This": [63], "minimalist": [64], "approach": [65], "produces": [66], "achieve": [69], "which": [72], "is": [73, 91], "competitive": [74], "with,": [75], "and": [76, 119, 130, 159, 177, 207, 217], "sometimes": [77], "surpasses,": [78], "SOTA": [79, 105], "RL": [80, 147], "built": [83], "upon": [84], "the": [85, 98, 109, 142, 184, 187, 198], "same": [86], "Crucially,": [89], "at": [93, 125], "fraction": [96], "computational": [99], "post-training": [100, 129], "cost": [101, 132, 137], "employed": [102], "existing": [104], "models.": [106], "In": [107, 203], "fact,": [108], "best": [110], "model": [112, 185, 218], "achieves": [113], ">20\\%": [115], "increase": [118], "43.33\\%": [120], "Pass@1": [121], "accuracy": [122], "on": [123], "AIME24,": [124], "\\$9": [127], "USD": [128], "evaluation": [131], "(i.e.,": [133], "estimated": [135], "260x": [136], "reduction).": [138], "Our": [139], "work": [140], "reveals": [141], "surprising": [143], "effectiveness": [144, 176], "efficient": [146], "via": [149], "LoRA.": [150], "We": [151], "validate": [152], "across": [154], "multiple": [155], "open-source": [156, 212], "datasets": [158], "various": [160], "ablation": [161], "settings": [162], "starting": [163], "single,": [166], "fixed": [167], "set": [168], "hyperparameters.": [170], "Furthermore,": [171], "hypothesize": [173], "efficiency": [178], "stem": [179], "from": [180], "LoRA": [181], "rapidly": [182], "adapting": [183], "structural": [188], "format": [189], "rewarded": [192], "RL,": [194], "while": [195], "largely": [196], "preserving": [197], "model's": [200], "underlying": [201], "knowledge.": [202], "service": [204], "accessibility": [206], "open": [208], "research,": [209], "fully": [211], "all": [213], "code,": [214], "training": [215], "logs,": [216], "weights": [219], "\\&": [220], "checkpoints.": [221]}, "primary_topic": {"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.6240000128746033, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.6240000128746033, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T11307", "display_name": "Domain Adaptation and Few-Shot Learning", "score": 0.1656000018119812, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10181", "display_name": "Natural Language Processing Techniques", "score": 0.06669999659061432, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "pmh:oai:arXiv.org:2504.15777", "is_oa": true, "landing_page_url": "http://arxiv.org/abs/2504.15777", "pdf_url": "https://arxiv.org/pdf/2504.15777", "source": {"id": "https://openalex.org/S4306400194", "display_name": "arXiv (Cornell University)", "issn_l": "2331-8422", "issn": ["2331-8422"], "is_oa": true, "is_in_doaj": false, "is_core": false, "listed_in": [], "host_organization": "https://openalex.org/I205783295", "host_organization_name": "Cornell University", "host_organization_lineage": ["https://openalex.org/I205783295"], "host_organization_lineage_names": [], "type": "repository"}, "license": "cc-by", "license_id": "https://openalex.org/licenses/cc-by", "version": "submittedVersion", "is_accepted": false, "is_published": false, "raw_source_name": null, "raw_type": "text"}, "cited_by_count": 0, "is_retracted": false}, {"id": "https://openalex.org/W4416716572", "doi": "https://doi.org/10.48550/arxiv.2511.11500", "title": "Honesty over Accuracy: Trustworthy Language Models through Reinforced Hesitation", "publication_year": 2025, "publication_date": "2025-11-14", "type": "preprint", "language": null, "authorships": [{"author_position": "first", "author": {"id": "https://openalex.org/A5041640864", "display_name": "Mohamad Amin Mohamadi", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Mohamadi, Mohamad Amin", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5100610986", "display_name": "Tianhao Wang", "orcid": "https://orcid.org/0000-0002-9017-7947", "observed_orcids": ["https://orcid.org/0000-0002-9017-7947"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Wang, Tianhao", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "last", "author": {"id": "https://openalex.org/A5100337924", "display_name": "Zhiyuan Li", "orcid": "https://orcid.org/0009-0003-5549-0721", "observed_orcids": ["https://orcid.org/0009-0003-5549-0721", "https://orcid.org/0000-0001-8446-0319", "https://orcid.org/0009-0003-3876-3214"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Li, Zhiyuan", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}], "abstract_inverted_index": {"Modern": [0], "language": [1], "models": [2, 22, 42, 106, 151, 193], "fail": [3], "a": [4, 66, 73, 108, 144, 177, 189], "fundamental": [5], "requirement": [6], "of": [7, 49, 93], "trustworthy": [8], "intelligence:": [9], "knowing": [10], "when": [11, 27], "not": [12], "to": [13, 75, 82, 194], "answer.": [14, 64], "Despite": [15], "achieving": [16], "impressive": [17], "accuracy": [18], "on": [19, 35, 97, 162], "benchmarks,": [20], "these": [21], "produce": [23, 126], "confident": [24], "hallucinations,": [25], "even": [26], "wrong": [28], "answers": [29], "carry": [30], "catastrophic": [31], "consequences.": [32], "Our": [33], "evaluations": [34], "GSM8K,": [36], "MedQA": [37], "and": [38], "GPQA": [39], "show": [40], "frontier": [41], "almost": [43], "never": [44], "abstain": [45], "despite": [46], "explicit": [47], "warnings": [48], "severe": [50], "penalties,": [51], "suggesting": [52], "that": [53, 58, 101, 139, 181], "prompts": [54], "cannot": [55], "override": [56], "training": [57, 113, 179], "rewards": [59, 85], "any": [60], "answer": [61], "over": [62], "no": [63], "As": [65], "remedy,": [67], "we": [68], "propose": [69], "Reinforced": [70], "Hesitation": [71], "(RH):": [72], "modification": [74], "Reinforcement": [76], "Learning": [77], "from": [78, 186], "Verifiable": [79], "Rewards": [80], "(RLVR)": [81], "use": [83], "ternary": [84], "(+1": [86], "correct,": [87], "0": [88], "abstention,": [89], "-$\u03bb$": [90], "error)": [91], "instead": [92], "binary.": [94], "Controlled": [95], "experiments": [96], "logic": [98], "puzzles": [99], "reveal": [100], "varying": [102], "$\u03bb$": [103], "produces": [104], "distinct": [105], "along": [107], "Pareto": [109], "frontier,": [110], "where": [111], "each": [112], "penalty": [114], "yields": [115], "the": [116, 159], "optimal": [117], "model": [118, 161], "for": [119], "its": [120], "corresponding": [121], "risk": [122, 154], "regime:": [123], "low": [124], "penalties": [125, 130], "aggressive": [127], "answerers,": [128], "high": [129], "conservative": [131], "abstainers.": [132], "We": [133], "then": [134], "introduce": [135], "two": [136], "inference": [137], "strategies": [138], "exploit": [140], "trained": [141], "abstention": [142, 175], "as": [143, 176], "coordination": [145, 190], "signal:": [146], "cascading": [147], "routes": [148], "queries": [149], "through": [150, 197], "with": [152, 168], "decreasing": [153], "tolerance,": [155], "while": [156], "self-cascading": [157], "re-queries": [158], "same": [160], "abstention.": [163], "Both": [164], "outperform": [165], "majority": [166], "voting": [167], "lower": [169], "computational": [170], "cost.": [171], "These": [172], "results": [173], "establish": [174], "first-class": [178], "objective": [180], "transforms": [182], "``I": [183], "don't": [184], "know''": [185], "failure": [187], "into": [188], "signal,": [191], "enabling": [192], "earn": [195], "trust": [196], "calibrated": [198], "honesty": [199], "about": [200], "their": [201], "limits.": [202]}, "primary_topic": {"id": "https://openalex.org/T11689", "display_name": "Adversarial Robustness in Machine Learning", "score": 0.4652999937534332, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T11689", "display_name": "Adversarial Robustness in Machine Learning", "score": 0.4652999937534332, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.25099998712539673, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10181", "display_name": "Natural Language Processing Techniques", "score": 0.09690000116825104, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "pmh:oai:arXiv.org:2511.11500", "is_oa": true, "landing_page_url": "http://arxiv.org/abs/2511.11500", "pdf_url": "https://arxiv.org/pdf/2511.11500", "source": {"id": "https://openalex.org/S4306400194", "display_name": "arXiv (Cornell University)", "issn_l": "2331-8422", "issn": ["2331-8422"], "is_oa": true, "is_in_doaj": false, "is_core": false, "listed_in": [], "host_organization": "https://openalex.org/I205783295", "host_organization_name": "Cornell University", "host_organization_lineage": ["https://openalex.org/I205783295"], "host_organization_lineage_names": [], "type": "repository"}, "license": null, "license_id": null, "version": "submittedVersion", "is_accepted": false, "is_published": false, "raw_source_name": null, "raw_type": "text"}, "cited_by_count": 0, "is_retracted": false}, {"id": "https://openalex.org/W4407764421", "doi": "https://doi.org/10.48550/arxiv.2502.13604", "title": "BeamLoRA: Beam-Constraint Low-Rank Adaptation", "publication_year": 2025, "publication_date": "2025-02-19", "type": "preprint", "language": "en", "authorships": [{"author_position": "first", "author": {"id": null, "display_name": "Gu, Naibin", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Gu, Naibin", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5101902940", "display_name": "Zhenyu Zhang", "orcid": "https://orcid.org/0009-0005-9932-0067", "observed_orcids": ["https://orcid.org/0009-0005-9932-0067", "https://orcid.org/0000-0003-2264-4205"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Zhang, Zhenyu", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5102728089", "display_name": "Xiyu Liu", "orcid": "https://orcid.org/0009-0003-5826-5431", "observed_orcids": ["https://orcid.org/0009-0003-5826-5431"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Liu, Xiyu", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5001781730", "display_name": "Peng Fu", "orcid": "https://orcid.org/0000-0001-9899-8566", "observed_orcids": ["https://orcid.org/0000-0001-9899-8566"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Fu, Peng", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5089094818", "display_name": "Zheng Lin", "orcid": "https://orcid.org/0000-0002-4463-5652", "observed_orcids": ["https://orcid.org/0000-0002-4463-5652", "https://orcid.org/0009-0006-4597-8557"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Lin, Zheng", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5101351951", "display_name": "Shuohuan Wang", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Wang, Shuohuan", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5077454145", "display_name": "Yuzhou Sun", "orcid": "https://orcid.org/0000-0001-9167-3398", "observed_orcids": ["https://orcid.org/0000-0001-9167-3398"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Sun, Yu", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": null, "display_name": "Wu, Hua", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Wu, Hua", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": null, "display_name": "Wang, Weiping", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Wang, Weiping", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "last", "author": {"id": "https://openalex.org/A5100386401", "display_name": "Haifeng Wang", "orcid": "https://orcid.org/0000-0002-3771-4504", "observed_orcids": ["https://orcid.org/0000-0002-3771-4504"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Wang, Haifeng", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}], "abstract_inverted_index": {"Due": [0], "to": [1, 45, 104], "the": [2, 21, 47, 59, 74, 80, 109, 116, 127, 161, 166], "demand": [3], "for": [4, 35, 115, 130], "efficient": [5], "fine-tuning": [6, 25, 75, 110], "of": [7, 20, 49, 67, 82, 163], "large": [8], "language": [9], "models,": [10], "Low-Rank": [11], "Adaptation": [12], "(LoRA)": [13], "has": [14], "been": [15], "widely": [16], "adopted": [17], "as": [18, 96], "one": [19], "most": [22], "effective": [23], "parameter-efficient": [24], "methods.": [26, 169], "Nevertheless,": [27], "while": [28, 125], "LoRA": [29, 50, 60, 94], "improves": [30], "efficiency,": [31], "there": [32], "remains": [33], "room": [34], "improvement": [36], "in": [37], "accuracy.": [38], "Herein,": [39], "we": [40, 88], "adopt": [41], "a": [42, 97, 105, 113, 136], "novel": [43], "perspective": [44], "assess": [46], "characteristics": [48], "ranks.": [51], "The": [52], "results": [53], "reveal": [54], "that": [55, 157], "different": [56], "ranks": [57], "within": [58], "modules": [61], "not": [62], "only": [63], "exhibit": [64], "varying": [65], "levels": [66], "importance": [68], "but": [69], "also": [70], "evolve": [71], "dynamically": [72, 121], "throughout": [73], "process,": [76], "which": [77, 91], "may": [78], "limit": [79], "performance": [81, 134, 162], "LoRA.": [83], "Based": [84], "on": [85], "these": [86], "findings,": [87], "propose": [89], "BeamLoRA,": [90], "conceptualizes": [92], "each": [93, 100], "module": [95], "beam": [98], "where": [99], "rank": [101], "naturally": [102], "corresponds": [103], "potential": [106], "sub-solution,": [107], "and": [108, 145, 153], "process": [111], "becomes": [112], "search": [114], "optimal": [117], "sub-solution": [118], "combination.": [119], "BeamLoRA": [120, 158], "eliminates": [122], "underperforming": [123], "sub-solutions": [124], "expanding": [126], "parameter": [128], "space": [129], "promising": [131], "ones,": [132], "enhancing": [133], "with": [135], "fixed": [137], "rank.": [138], "Extensive": [139], "experiments": [140], "across": [141], "three": [142], "base": [143], "models": [144], "12": [146], "datasets": [147], "spanning": [148], "math": [149], "reasoning,": [150], "code": [151], "generation,": [152], "commonsense": [154], "reasoning": [155], "demonstrate": [156], "consistently": [159], "enhances": [160], "LoRA,": [164], "surpassing": [165], "other": [167], "baseline": [168]}, "primary_topic": {"id": "https://openalex.org/T11307", "display_name": "Domain Adaptation and Few-Shot Learning", "score": 0.7490000128746033, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T11307", "display_name": "Domain Adaptation and Few-Shot Learning", "score": 0.7490000128746033, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.21809999644756317, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10181", "display_name": "Natural Language Processing Techniques", "score": 0.009600000455975533, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "pmh:oai:arXiv.org:2502.13604", "is_oa": true, "landing_page_url": "http://arxiv.org/abs/2502.13604", "pdf_url": "https://arxiv.org/pdf/2502.13604", "source": {"id": "https://openalex.org/S4306400194", "display_name": "arXiv (Cornell University)", "issn_l": "2331-8422", "issn": ["2331-8422"], "is_oa": true, "is_in_doaj": false, "is_core": false, "listed_in": [], "host_organization": "https://openalex.org/I205783295", "host_organization_name": "Cornell University", "host_organization_lineage": ["https://openalex.org/I205783295"], "host_organization_lineage_names": [], "type": "repository"}, "license": null, "license_id": null, "version": "submittedVersion", "is_accepted": false, "is_published": false, "raw_source_name": null, "raw_type": "text"}, "cited_by_count": 0, "is_retracted": false}, {"id": "https://openalex.org/W4407759571", "doi": "https://doi.org/10.48550/arxiv.2502.12623", "title": "DeepResonance: Enhancing Multimodal Music Understanding via Music-centric Multi-way Instruction Tuning", "publication_year": 2025, "publication_date": "2025-02-18", "type": "preprint", "language": "en", "authorships": [{"author_position": "first", "author": {"id": "https://openalex.org/A5017982536", "display_name": "Zhuoyuan Mao", "orcid": "https://orcid.org/0000-0001-5273-2738", "observed_orcids": ["https://orcid.org/0000-0001-5273-2738"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Mao, Zhuoyuan", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5101793596", "display_name": "Mengjie Zhao", "orcid": "https://orcid.org/0009-0009-2391-4028", "observed_orcids": ["https://orcid.org/0009-0009-2391-4028"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Zhao, Mengjie", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5110490268", "display_name": "Qiyu Wu", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Wu, Qiyu", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5014478861", "display_name": "Hiromi Wakaki", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Wakaki, Hiromi", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "last", "author": {"id": "https://openalex.org/A5088754502", "display_name": "Yuki Mitsufuji", "orcid": "https://orcid.org/0000-0002-6806-6140", "observed_orcids": ["https://orcid.org/0000-0002-6806-6140"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Mitsufuji, Yuki", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}], "abstract_inverted_index": {"Recent": [0], "advancements": [1], "in": [2], "music": [3, 11, 33, 50, 54, 67, 109, 145], "large": [4], "language": [5], "models": [6, 165], "(LLMs)": [7], "have": [8], "significantly": [9], "improved": [10], "understanding": [12, 55, 68, 146], "tasks,": [13, 147], "which": [14], "involve": [15], "the": [16, 38, 149, 152, 156, 163], "model's": [17], "ability": [18], "to": [19, 52, 100, 103, 123, 128], "analyze": [20], "and": [21, 34, 48, 81, 91, 96, 107, 118, 155, 166], "interpret": [22], "various": [23], "musical": [24], "elements.": [25], "These": [26], "improvements": [27], "primarily": [28], "focused": [29], "on": [30], "integrating": [31], "both": [32, 105], "text": [35, 131], "inputs.": [36], "However,": [37], "potential": [39], "of": [40, 151, 159], "incorporating": [41], "additional": [42], "modalities": [43, 154], "such": [44], "as": [45], "images,": [46], "videos": [47], "textual": [49, 108], "features": [51], "enhance": [53, 124], "remains": [56], "unexplored.": [57], "To": [58, 84], "bridge": [59], "this": [60, 85], "gap,": [61], "we": [62, 87, 168], "propose": [63], "DeepResonance,": [64], "a": [65, 119], "multimodal": [66], "LLM": [69], "fine-tuned": [70], "via": [71], "multi-way": [72, 76, 135], "instruction": [73, 136], "tuning": [74], "with": [75], "aligned": [77], "music,": [78], "text,": [79], "image,": [80], "video": [82], "data.": [83], "end,": [86], "construct": [88], "Music4way-MI2T,": [89], "Music4way-MV2T,": [90], "Music4way-Any2T,": [92], "three": [93], "4-way": [94], "training": [95], "evaluation": [97], "datasets": [98, 167], "designed": [99], "enable": [101], "DeepResonance": [102], "integrate": [104], "visual": [106], "feature": [110], "content.": [111], "We": [112, 161], "also": [113], "introduce": [114], "multi-sampled": [115], "ImageBind": [116], "embeddings": [117], "pre-LLM": [120], "fusion": [121, 126], "Transformer": [122], "modality": [125], "prior": [127], "input": [129], "into": [130], "LLMs,": [132], "tailoring": [133], "for": [134], "tuning.": [137], "Our": [138], "model": [139], "achieves": [140], "state-of-the-art": [141], "performances": [142], "across": [143], "six": [144], "highlighting": [148], "benefits": [150], "auxiliary": [153], "structural": [157], "superiority": [158], "DeepResonance.": [160], "open-source": [162], "codes,": [164], "constructed:": [169], "github.com/sony/DeepResonance.": [170]}, "primary_topic": {"id": "https://openalex.org/T11309", "display_name": "Music and Audio Processing", "score": 0.5702999830245972, "subfield": {"id": "https://openalex.org/subfields/1711", "display_name": "Signal Processing"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T11309", "display_name": "Music and Audio Processing", "score": 0.5702999830245972, "subfield": {"id": "https://openalex.org/subfields/1711", "display_name": "Signal Processing"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T11714", "display_name": "Multimodal Machine Learning Applications", "score": 0.40869998931884766, "subfield": {"id": "https://openalex.org/subfields/1707", "display_name": "Computer Vision and Pattern Recognition"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.006500000134110451, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "pmh:oai:arXiv.org:2502.12623", "is_oa": true, "landing_page_url": "http://arxiv.org/abs/2502.12623", "pdf_url": "https://arxiv.org/pdf/2502.12623", "source": {"id": "https://openalex.org/S4306400194", "display_name": "arXiv (Cornell University)", "issn_l": "2331-8422", "issn": ["2331-8422"], "is_oa": true, "is_in_doaj": false, "is_core": false, "listed_in": [], "host_organization": "https://openalex.org/I205783295", "host_organization_name": "Cornell University", "host_organization_lineage": ["https://openalex.org/I205783295"], "host_organization_lineage_names": [], "type": "repository"}, "license": null, "license_id": null, "version": "submittedVersion", "is_accepted": false, "is_published": false, "raw_source_name": null, "raw_type": "text"}, "cited_by_count": 0, "is_retracted": false}, {"id": "https://openalex.org/W4414630937", "doi": "https://doi.org/10.1109/taslpro.2025.3612842", "title": "Privacy-Preserving Parameter-Efficient Fine-Tuning for Large Language Model Services", "publication_year": 2025, "publication_date": "2025-01-01", "type": "article", "language": "en", "authorships": [{"author_position": "first", "author": {"id": "https://openalex.org/A5100762830", "display_name": "Yansong Li", "orcid": "https://orcid.org/0009-0005-8542-1190", "observed_orcids": ["https://orcid.org/0009-0005-8542-1190", "https://orcid.org/0009-0004-4694-8419"]}, "institutions": [{"id": "https://openalex.org/I153718931", "display_name": "University of Ottawa", "ror": "https://ror.org/03c4mmv16", "country_code": "CA", "type": "education", "lineage": ["https://openalex.org/I153718931"]}], "countries": ["CA"], "is_corresponding": false, "raw_author_name": "Yansong Li", "raw_affiliation_strings": ["University of Ottawa, Ottawa, ON, Canada", "University of Ottawa, Ottawa, Canada"], "raw_orcid": "https://orcid.org/0009-0004-4694-8419", "affiliations": [{"raw_affiliation_string": "University of Ottawa, Ottawa, ON, Canada", "institution_ids": ["https://openalex.org/I153718931"]}, {"raw_affiliation_string": "University of Ottawa, Ottawa, Canada", "institution_ids": ["https://openalex.org/I153718931"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5086583247", "display_name": "Zhixing Tan", "orcid": "https://orcid.org/0000-0002-2426-6220", "observed_orcids": ["https://orcid.org/0000-0002-2426-6220"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Zhixing Tan", "raw_affiliation_strings": ["Zhongguancun Laboratory, Beijing, China"], "raw_orcid": "https://orcid.org/0000-0002-2426-6220", "affiliations": [{"raw_affiliation_string": "Zhongguancun Laboratory, Beijing, China", "institution_ids": []}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5043554078", "display_name": "Paula Branco", "orcid": "https://orcid.org/0000-0002-9917-3694", "observed_orcids": ["https://orcid.org/0000-0002-9917-3694"]}, "institutions": [{"id": "https://openalex.org/I153718931", "display_name": "University of Ottawa", "ror": "https://ror.org/03c4mmv16", "country_code": "CA", "type": "education", "lineage": ["https://openalex.org/I153718931"]}], "countries": ["CA"], "is_corresponding": false, "raw_author_name": "Paula Branco", "raw_affiliation_strings": ["School of Electrical Engineering and Computer Science, University of Ottawa, Ottawa, ON, Canada", "School of Electrical Engineering and Computer Science, University of Ottawa, Ottawa, Canada"], "raw_orcid": "https://orcid.org/0000-0002-9917-3694", "affiliations": [{"raw_affiliation_string": "School of Electrical Engineering and Computer Science, University of Ottawa, Ottawa, ON, Canada", "institution_ids": ["https://openalex.org/I153718931"]}, {"raw_affiliation_string": "School of Electrical Engineering and Computer Science, University of Ottawa, Ottawa, Canada", "institution_ids": ["https://openalex.org/I153718931"]}]}, {"author_position": "last", "author": {"id": "https://openalex.org/A5100749600", "display_name": "Yang Liu", "orcid": "https://orcid.org/0009-0005-7193-0795", "observed_orcids": ["https://orcid.org/0009-0005-7193-0795", "https://orcid.org/0000-0002-3087-242X", "https://orcid.org/0000-0002-5463-0791"]}, "institutions": [{"id": "https://openalex.org/I99065089", "display_name": "Tsinghua University", "ror": "https://ror.org/03cve4549", "country_code": "CN", "type": "education", "lineage": ["https://openalex.org/I99065089"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Yang Liu", "raw_affiliation_strings": ["Department of Computer Science and Technology, Tsinghua University, Beijing, China"], "raw_orcid": "https://orcid.org/0000-0002-3087-242X", "affiliations": [{"raw_affiliation_string": "Department of Computer Science and Technology, Tsinghua University, Beijing, China", "institution_ids": ["https://openalex.org/I99065089"]}]}], "abstract_inverted_index": {"Parameter-Efficient": [0, 61], "Fine-Tuning": [1, 62], "(PEFT)": [2], "provides": [3], "a": [4, 64, 75, 87, 105], "practical": [5], "way": [6], "for": [7, 70], "users": [8, 80], "to": [9, 42, 81, 121], "customize": [10], "Large": [11], "Language": [12], "Models": [13], "(LLMs)": [14], "with": [15, 51, 115], "their": [16, 83], "private": [17, 29], "data": [18, 30, 44, 84], "in": [19], "LLM": [20, 40, 71], "service": [21], "scenarios.": [22], "However,": [23], "the": [24, 37, 116, 127], "inherently": [25], "sensitive": [26], "nature": [27], "of": [28, 39, 129], "demands": [31], "robust": [32], "privacy": [33, 68, 77, 91, 143], "preservation": [34], "measures": [35], "during": [36], "customization": [38], "services": [41], "ensure": [43], "security,": [45], "maintain": [46], "user": [47], "trust,": [48], "and": [49], "comply": [50], "stringent": [52], "regulatory": [53], "standards.": [54], "Based": [55], "on": [56, 100], "PEFT,": [57], "we": [58, 103], "propose": [59], "Privacy-Preserving": [60], "(RAPT),": [63], "framework": [65], "that": [66, 111, 134], "offers": [67], "protection": [69], "services.": [72], "RAPT": [73, 135], "adopts": [74], "local": [76, 89], "approach,": [78], "enabling": [79], "privatize": [82], "locally": [85], "using": [86], "text-to-text": [88], "differential": [90], "mechanism.": [92], "Since": [93], "PEFT": [94], "performs": [95], "poorly": [96], "when": [97], "directly": [98], "trained": [99, 113], "privatized": [101, 107], "data,": [102], "introduce": [104], "novel": [106], "token": [108], "reconstruction": [109], "task": [110], "is": [112], "jointly": [114], "downstream": [117], "task,": [118], "allowing": [119], "LLMs": [120], "learn": [122], "better": [123], "task-dependent": [124], "representations.": [125], "Despite": [126], "simplicity": [128], "our": [130], "framework,": [131], "experiments": [132], "show": [133], "achieves": [136], "competitive": [137], "performance": [138], "across": [139], "tasks": [140], "while": [141], "providing": [142], "guarantees": [144], "against": [145], "adversaries.": [146]}, "primary_topic": {"id": "https://openalex.org/T10764", "display_name": "Privacy-Preserving Technologies in Data", "score": 0.9990000128746033, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T10764", "display_name": "Privacy-Preserving Technologies in Data", "score": 0.9990000128746033, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.00019999999494757503, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T11689", "display_name": "Adversarial Robustness in Machine Learning", "score": 0.00019999999494757503, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "doi:10.1109/taslpro.2025.3612842", "is_oa": false, "landing_page_url": "https://doi.org/10.1109/taslpro.2025.3612842", "pdf_url": null, "source": {"id": "https://openalex.org/S5407042752", "display_name": "IEEE Transactions on Audio Speech and Language Processing", "issn_l": "2998-4173", "issn": ["2998-4173"], "is_oa": false, "is_in_doaj": false, "is_core": false, "listed_in": ["jufo-3"], "host_organization": null, "host_organization_name": null, "host_organization_lineage": [], "host_organization_lineage_names": [], "type": "journal"}, "license": null, "license_id": null, "version": "publishedVersion", "is_accepted": true, "is_published": true, "raw_source_name": "IEEE Transactions on Audio, Speech and Language Processing", "raw_type": "journal-article"}, "cited_by_count": 19, "is_retracted": false}, {"id": "https://openalex.org/W4417300694", "doi": "https://doi.org/10.48550/arxiv.2505.12435", "title": "SGDPO: Self-Guided Direct Preference Optimization for Language Model Alignment", "publication_year": 2025, "publication_date": "2025-05-18", "type": "preprint", "language": "en", "authorships": [{"author_position": "first", "author": {"id": "https://openalex.org/A5110106146", "display_name": "Wenqiao Zhu", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Zhu, Wenqiao", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5100689701", "display_name": "Liu Jing Ji", "orcid": "https://orcid.org/0000-0002-5386-3550", "observed_orcids": ["https://orcid.org/0000-0002-5386-3550"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Liu, Ji", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5100444637", "display_name": "Lulu Wang", "orcid": "https://orcid.org/0000-0002-6796-2533", "observed_orcids": ["https://orcid.org/0000-0002-6796-2533"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Wang, Lulu", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5100341520", "display_name": "Jun Wu", "orcid": "https://orcid.org/0009-0002-0730-8860", "observed_orcids": ["https://orcid.org/0009-0002-0730-8860"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Wu, Jun", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "last", "author": {"id": "https://openalex.org/A5074865219", "display_name": "Yulun Zhang", "orcid": "https://orcid.org/0000-0002-2288-5079", "observed_orcids": ["https://orcid.org/0000-0002-2288-5079"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Zhang, Yulun", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}], "abstract_inverted_index": {"Direct": [0, 59], "Preference": [1, 60], "Optimization": [2, 61], "(DPO)": [3], "is": [4, 36], "broadly": [5], "utilized": [6], "for": [7, 80], "aligning": [8], "Large": [9], "Language": [10], "Models": [11], "(LLMs)": [12], "with": [13], "human": [14], "values": [15], "because": [16], "of": [17, 30, 41, 86, 97, 135], "its": [18, 21, 103], "flexibility.": [19], "Despite": [20], "effectiveness,": [22], "it": [23], "has": [24], "been": [25], "observed": [26], "that": [27], "the": [28, 39, 72, 76, 84, 121, 124, 133], "capability": [29], "DPO": [31, 42], "to": [32, 70, 140], "generate": [33], "human-preferred": [34], "response": [35], "limited": [37], "and": [38, 88, 101, 114, 127, 131], "results": [40, 119, 126], "are": [43], "far": [44], "from": [45], "resilient.": [46], "To": [47], "address": [48], "these": [49], "limitations,": [50], "in": [51], "this": [52], "paper": [53], "we": [54, 107], "propose": [55], "a": [56, 67, 93], "novel": [57], "Self-Guided": [58], "algorithm,": [62], "i.e.,": [63], "SGDPO,": [64], "which": [65], "incorporates": [66], "pilot": [68], "term": [69], "steer": [71], "gradient": [73], "flow": [74], "during": [75], "optimization": [77], "process,": [78], "allowing": [79], "fine-grained": [81], "control": [82], "over": [83], "updates": [85], "chosen": [87], "rejected": [89], "rewards.": [90], "We": [91], "provide": [92], "detailed": [94], "theoretical": [95, 129], "analysis": [96, 130], "our": [98, 128, 136], "proposed": [99, 137], "method": [100], "elucidate": [102], "operational": [104], "mechanism.": [105], "Furthermore,": [106], "conduct": [108], "comprehensive": [109], "experiments": [110], "on": [111], "various": [112], "models": [113], "benchmarks.": [115], "The": [116], "extensive": [117], "experimental": [118], "demonstrate": [120], "consistency": [122], "between": [123], "empirical": [125], "confirm": [132], "effectiveness": [134], "approach": [138], "(up": [139], "9.19%": [141], "higher": [142], "score).": [143]}, "primary_topic": {"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.33399999141693115, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.33399999141693115, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10181", "display_name": "Natural Language Processing Techniques", "score": 0.18490000069141388, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T11307", "display_name": "Domain Adaptation and Few-Shot Learning", "score": 0.0731000006198883, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "pmh:oai:arXiv.org:2505.12435", "is_oa": true, "landing_page_url": "http://arxiv.org/abs/2505.12435", "pdf_url": "https://arxiv.org/pdf/2505.12435", "source": {"id": "https://openalex.org/S4306400194", "display_name": "arXiv (Cornell University)", "issn_l": "2331-8422", "issn": ["2331-8422"], "is_oa": true, "is_in_doaj": false, "is_core": false, "listed_in": [], "host_organization": "https://openalex.org/I205783295", "host_organization_name": "Cornell University", "host_organization_lineage": ["https://openalex.org/I205783295"], "host_organization_lineage_names": [], "type": "repository"}, "license": null, "license_id": null, "version": "submittedVersion", "is_accepted": false, "is_published": false, "raw_source_name": null, "raw_type": "text"}, "cited_by_count": 0, "is_retracted": false}, {"id": "https://openalex.org/W4417097679", "doi": "https://doi.org/10.48550/arxiv.2506.06093", "title": "Reinforcing Code Generation: Improving Text-to-SQL with Execution-Based Learning", "publication_year": 2025, "publication_date": "2025-06-06", "type": "preprint", "language": "en", "authorships": [{"author_position": "first", "author": {"id": "https://openalex.org/A5087448996", "display_name": "Atharv Kulkarni", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Kulkarni, Atharv", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "last", "author": {"id": "https://openalex.org/A5013135203", "display_name": "Vivek Srikumar", "orcid": "https://orcid.org/0000-0003-0419-6568", "observed_orcids": ["https://orcid.org/0000-0003-0419-6568"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Srikumar, Vivek", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}], "abstract_inverted_index": {"In": [0], "this": [1, 53], "work,": [2], "we": [3, 39], "study": [4], "the": [5, 61, 67, 70, 93, 96, 123, 130, 152, 156, 160, 167], "problem": [6, 54, 59], "of": [7, 30, 72, 125, 132, 169, 178], "code": [8, 136], "generation": [9], "with": [10, 16, 35, 47, 118], "a": [11, 17, 41, 48, 56, 85, 88, 105], "large": [12], "language": [13, 25], "model": [14, 42, 62, 133, 153], "(LLM),": [15], "focus": [18], "on": [19], "generating": [20], "SQL": [21, 135], "queries": [22], "from": [23, 66, 137, 145], "natural": [24], "questions.": [26], "We": [27, 51, 91, 103, 115], "ask:": [28], "Instead": [29], "using": [31, 170], "supervised": [32], "fine": [33], "tuning": [34], "text-code": [36], "pairs,": [37, 127], "can": [38], "tune": [40], "by": [43], "having": [44], "it": [45], "interact": [46], "database": [49], "engine?": [50], "frame": [52], "as": [55], "reinforcement": [57], "learning": [58], "where": [60], "receives": [63], "execution-based": [64, 171], "feedback": [65, 172], "environment": [68], "in": [69, 122], "form": [71, 124], "scalar": [73], "rewards.": [74], "These": [75], "rewards": [76, 94], "penalize": [77], "execution": [78], "failures": [79], "and": [80, 111], "assign": [81], "positive": [82], "values": [83], "when": [84], "query": [86], "returns": [87], "correct": [89], "answer.": [90], "use": [92, 104], "within": [95], "Group": [97], "Relative": [98], "Policy": [99], "Optimization": [100], "(GRPO)": [101], "framework.": [102], "tabular": [106], "reasoning": [107, 176], "benchmark": [108], "to": [109, 139, 147, 159, 173], "test": [110], "evaluate": [112], "our": [113], "findings.": [114], "find": [116], "that": [117], "only": [119], "weak": [120], "supervision": [121], "question-answer": [126], "RL-tuning": [128], "improves": [129], "accuracy": [131], "generated": [134], "31.49": [138], "49.83": [140], "while": [141], "reducing": [142], "error": [143], "percentage": [144], "25.43%": [146], "14.71%.": [148], "This": [149], "improvement": [150], "allowed": [151], "nearly": [154], "match": [155], "performance": [157, 158], "larger": [161], "SQLCoder-70B": [162], "model.": [163], "Our": [164], "work": [165], "demonstrates": [166], "potential": [168], "improve": [174], "symbolic": [175], "capabilities": [177], "LLMs.": [179]}, "primary_topic": {"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.3230000138282776, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.3230000138282776, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10260", "display_name": "Software Engineering Research", "score": 0.2345000058412552, "subfield": {"id": "https://openalex.org/subfields/1710", "display_name": "Information Systems"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10181", "display_name": "Natural Language Processing Techniques", "score": 0.21359999477863312, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "pmh:oai:arXiv.org:2506.06093", "is_oa": true, "landing_page_url": "http://arxiv.org/abs/2506.06093", "pdf_url": "https://arxiv.org/pdf/2506.06093", "source": {"id": "https://openalex.org/S4306400194", "display_name": "arXiv (Cornell University)", "issn_l": "2331-8422", "issn": ["2331-8422"], "is_oa": true, "is_in_doaj": false, "is_core": false, "listed_in": [], "host_organization": "https://openalex.org/I205783295", "host_organization_name": "Cornell University", "host_organization_lineage": ["https://openalex.org/I205783295"], "host_organization_lineage_names": [], "type": "repository"}, "license": null, "license_id": null, "version": "submittedVersion", "is_accepted": false, "is_published": false, "raw_source_name": null, "raw_type": "text"}, "cited_by_count": 1, "is_retracted": false}, {"id": "https://openalex.org/W4406784832", "doi": "https://doi.org/10.48550/arxiv.2501.11463", "title": "Curiosity-Driven Reinforcement Learning from Human Feedback", "publication_year": 2025, "publication_date": "2025-01-20", "type": "preprint", "language": "en", "authorships": [{"author_position": "first", "author": {"id": "https://openalex.org/A5101563679", "display_name": "Haoran Sun", "orcid": "https://orcid.org/0009-0007-7817-1773", "observed_orcids": ["https://orcid.org/0009-0007-7817-1773"]}, "institutions": [{"id": "https://openalex.org/I98301712", "display_name": "Baidu (China)", "ror": "https://ror.org/03vs3wt56", "country_code": "CN", "type": "company", "lineage": ["https://openalex.org/I98301712"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Sun, Haoran", "raw_affiliation_strings": ["Baidu Inc"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Baidu Inc", "institution_ids": ["https://openalex.org/I98301712"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5072082849", "display_name": "Yekun Chai", "orcid": null, "observed_orcids": []}, "institutions": [{"id": "https://openalex.org/I98301712", "display_name": "Baidu (China)", "ror": "https://ror.org/03vs3wt56", "country_code": "CN", "type": "company", "lineage": ["https://openalex.org/I98301712"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Chai, Yekun", "raw_affiliation_strings": ["Baidu Inc"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Baidu Inc", "institution_ids": ["https://openalex.org/I98301712"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5058880150", "display_name": "Shuohuan Wang", "orcid": null, "observed_orcids": []}, "institutions": [{"id": "https://openalex.org/I98301712", "display_name": "Baidu (China)", "ror": "https://ror.org/03vs3wt56", "country_code": "CN", "type": "company", "lineage": ["https://openalex.org/I98301712"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Wang, Shuohuan", "raw_affiliation_strings": ["Baidu Inc"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Baidu Inc", "institution_ids": ["https://openalex.org/I98301712"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5107072461", "display_name": "Yu Ming Sun", "orcid": "https://orcid.org/0000-0003-3525-2753", "observed_orcids": ["https://orcid.org/0000-0003-3525-2753"]}, "institutions": [{"id": "https://openalex.org/I98301712", "display_name": "Baidu (China)", "ror": "https://ror.org/03vs3wt56", "country_code": "CN", "type": "company", "lineage": ["https://openalex.org/I98301712"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Sun, Yu", "raw_affiliation_strings": ["Baidu Inc"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Baidu Inc", "institution_ids": ["https://openalex.org/I98301712"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5112985399", "display_name": "Hua Wu", "orcid": "https://orcid.org/0000-0001-8254-1561", "observed_orcids": ["https://orcid.org/0000-0001-8254-1561"]}, "institutions": [{"id": "https://openalex.org/I98301712", "display_name": "Baidu (China)", "ror": "https://ror.org/03vs3wt56", "country_code": "CN", "type": "company", "lineage": ["https://openalex.org/I98301712"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Wu, Hua", "raw_affiliation_strings": ["Baidu Inc"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Baidu Inc", "institution_ids": ["https://openalex.org/I98301712"]}]}, {"author_position": "last", "author": {"id": null, "display_name": "Wang, Haifeng", "orcid": null, "observed_orcids": []}, "institutions": [{"id": "https://openalex.org/I98301712", "display_name": "Baidu (China)", "ror": "https://ror.org/03vs3wt56", "country_code": "CN", "type": "company", "lineage": ["https://openalex.org/I98301712"]}], "countries": ["CN"], "is_corresponding": false, "raw_author_name": "Wang, Haifeng", "raw_affiliation_strings": ["Baidu Inc"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Baidu Inc", "institution_ids": ["https://openalex.org/I98301712"]}]}], "abstract_inverted_index": {"Reinforcement": [0], "learning": [1], "from": [2, 40], "human": [3, 16, 108], "feedback": [4], "(RLHF)": [5], "has": [6], "proven": [7], "effective": [8], "in": [9, 43, 98], "aligning": [10], "large": [11], "language": [12], "models": [13], "(LLMs)": [14], "with": [15, 107], "preferences,": [17], "but": [18], "often": [19], "at": [20, 120], "the": [21, 75], "cost": [22], "of": [23, 77, 85], "reduced": [24], "output": [25, 68], "diversity.": [26], "This": [27], "trade-off": [28], "between": [29], "diversity": [30, 69, 99], "and": [31, 70, 90], "alignment": [32, 71, 106], "quality": [33], "remains": [34], "a": [35, 51, 83], "significant": [36, 96], "challenge.": [37], "Drawing": [38], "inspiration": [39], "curiosity-driven": [41, 48], "exploration": [42], "reinforcement": [44], "learning,": [45], "we": [46], "introduce": [47], "RLHF": [49], "(CD-RLHF),": [50], "framework": [52], "that": [53], "incorporates": [54], "intrinsic": [55], "rewards": [56], "for": [57], "novel": [58], "states,": [59], "alongside": [60], "traditional": [61], "sparse": [62], "extrinsic": [63], "rewards,": [64], "to": [65, 111], "optimize": [66], "both": [67], "quality.": [72], "We": [73, 114], "demonstrate": [74], "effectiveness": [76], "CD-RLHF": [78], "through": [79], "extensive": [80], "experiments": [81], "on": [82, 100], "range": [84], "tasks,": [86], "including": [87], "text": [88], "summarization": [89], "instruction": [91], "following.": [92], "Our": [93], "approach": [94], "achieves": [95], "gains": [97], "multiple": [101], "diversity-oriented": [102], "metrics": [103], "while": [104], "maintaining": [105], "preferences": [109], "comparable": [110], "standard": [112], "RLHF.": [113], "make": [115], "our": [116], "code": [117], "publicly": [118], "available": [119], "https://github.com/ernie-research/CD-RLHF.": [121]}, "primary_topic": {"id": "https://openalex.org/T10462", "display_name": "Reinforcement Learning in Robotics", "score": 0.7860999703407288, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T10462", "display_name": "Reinforcement Learning in Robotics", "score": 0.7860999703407288, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.06930000334978104, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10181", "display_name": "Natural Language Processing Techniques", "score": 0.04349999874830246, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "pmh:oai:arXiv.org:2501.11463", "is_oa": true, "landing_page_url": "http://arxiv.org/abs/2501.11463", "pdf_url": "https://arxiv.org/pdf/2501.11463", "source": {"id": "https://openalex.org/S4306400194", "display_name": "arXiv (Cornell University)", "issn_l": "2331-8422", "issn": ["2331-8422"], "is_oa": true, "is_in_doaj": false, "is_core": false, "listed_in": [], "host_organization": "https://openalex.org/I205783295", "host_organization_name": "Cornell University", "host_organization_lineage": ["https://openalex.org/I205783295"], "host_organization_lineage_names": [], "type": "repository"}, "license": null, "license_id": null, "version": "submittedVersion", "is_accepted": false, "is_published": false, "raw_source_name": null, "raw_type": "text"}, "cited_by_count": 1, "is_retracted": false}, {"id": "https://openalex.org/W4416269536", "doi": "https://doi.org/10.48550/arxiv.2511.04869", "title": "Trained on Tokens, Calibrated on Concepts: The Emergence of Semantic Calibration in LLMs", "publication_year": 2025, "publication_date": "2025-11-06", "type": "preprint", "language": null, "authorships": [{"author_position": "first", "author": {"id": "https://openalex.org/A5110716768", "display_name": "Preetum Nakkiran", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Nakkiran, Preetum", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5076803200", "display_name": "Arwen Bradley", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Bradley, Arwen", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5050308555", "display_name": "Adam Goli\u0144ski", "orcid": "https://orcid.org/0000-0001-8603-1171", "observed_orcids": ["https://orcid.org/0000-0001-8603-1171"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Goli\u0144ski, Adam", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5021079415", "display_name": "Eug\u00e8ne Ndiaye", "orcid": "https://orcid.org/0000-0002-4898-434X", "observed_orcids": ["https://orcid.org/0000-0002-4898-434X"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Ndiaye, Eugene", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5035574656", "display_name": "Michael Kirchhof", "orcid": "https://orcid.org/0000-0003-4521-9391", "observed_orcids": ["https://orcid.org/0000-0003-4521-9391"]}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Kirchhof, Michael", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}, {"author_position": "last", "author": {"id": null, "display_name": "Williamson, Sinead", "orcid": null, "observed_orcids": []}, "institutions": [], "countries": [], "is_corresponding": false, "raw_author_name": "Williamson, Sinead", "raw_affiliation_strings": [], "raw_orcid": null, "affiliations": []}], "abstract_inverted_index": {"Large": [0], "Language": [1], "Models": [2], "(LLMs)": [3], "often": [4], "lack": [5], "meaningful": [6], "confidence": [7, 28, 61], "estimates": [8], "for": [9, 81], "their": [10, 34, 146], "outputs.": [11], "While": [12], "base": [13, 52, 135], "LLMs": [14, 53, 136, 171], "are": [15, 54, 172], "known": [16], "to": [17, 71, 131], "exhibit": [18], "next-token": [19, 90], "calibration,": [20, 51, 184], "it": [21], "remains": [22], "unclear": [23], "whether": [24], "they": [25, 57, 142], "can": [26, 58, 143], "assess": [27, 60], "in": [29, 62, 208], "the": [30, 37, 197], "actual": [31], "meaning": [32], "of": [33, 49, 89, 109, 115, 121, 161, 201], "responses": [35], "beyond": [36], "token": [38], "level.": [39], "We": [40, 157], "find": [41], "that,": [42], "when": [43, 141, 202], "using": [44], "a": [45, 79, 87, 93, 106, 113, 119, 132, 155], "certain": [46], "sampling-based": [47], "notion": [48, 114], "semantic": [50, 83, 150, 205], "remarkably": [55], "well-calibrated:": [56], "meaningfully": [59], "open-domain": [63], "question-answering": [64, 176], "tasks,": [65, 177], "despite": [66], "not": [67], "being": [68], "explicitly": [69], "trained": [70], "do": [72], "so.": [73], "Our": [74], "main": [75], "theoretical": [76, 128], "contribution": [77], "establishes": [78], "mechanism": [80, 129], "why": [82, 204], "calibration": [84, 97, 116, 206], "emerges": [85, 207], "as": [86], "byproduct": [88], "prediction,": [91, 163], "leveraging": [92], "recent": [94], "connection": [95], "between": [96], "and": [98, 185, 203], "local": [99], "loss": [100], "optimality.": [101], "The": [102], "theory": [103], "relies": [104], "on": [105], "general": [107], "definition": [108], "\"B-calibration,\"": [110], "which": [111, 164], "is": [112], "parameterized": [117], "by": [118], "choice": [120], "equivalence": [122], "classes": [123, 152], "(semantic": [124], "or": [125], "otherwise).": [126], "This": [127], "leads": [130], "testable": [133], "prediction:": [134], "will": [137], "be": [138], "semantically": [139, 173], "calibrated": [140, 174], "easily": [144], "predict": [145], "own": [147], "distribution": [148], "over": [149], "answer": [151], "before": [153], "generating": [154], "response.": [156], "state": [158], "three": [159], "implications": [160], "this": [162, 183], "we": [165], "validate": [166], "through": [167], "experiments:": [168], "(1)": [169], "Base": [170], "across": [175], "(2)": [178], "RL": [179], "instruction-tuning": [180], "systematically": [181], "breaks": [182, 189], "(3)": [186], "chain-of-thought": [187], "reasoning": [188], "calibration.": [190], "To": [191], "our": [192, 194], "knowledge,": [193], "work": [195], "provides": [196], "first": [198], "principled": [199], "explanation": [200], "LLMs.": [209]}, "primary_topic": {"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.5967000126838684, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, "topics": [{"id": "https://openalex.org/T10028", "display_name": "Topic Modeling", "score": 0.5967000126838684, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T11689", "display_name": "Adversarial Robustness in Machine Learning", "score": 0.23839999735355377, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}, {"id": "https://openalex.org/T10181", "display_name": "Natural Language Processing Techniques", "score": 0.04650000110268593, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "pmh:oai:arXiv.org:2511.04869", "is_oa": true, "landing_page_url": "http://arxiv.org/abs/2511.04869", "pdf_url": "https://arxiv.org/pdf/2511.04869", "source": {"id": "https://openalex.org/S4306400194", "display_name": "arXiv (Cornell University)", "issn_l": "2331-8422", "issn": ["2331-8422"], "is_oa": true, "is_in_doaj": false, "is_core": false, "listed_in": [], "host_organization": "https://openalex.org/I205783295", "host_organization_name": "Cornell University", "host_organization_lineage": ["https://openalex.org/I205783295"], "host_organization_lineage_names": [], "type": "repository"}, "license": null, "license_id": null, "version": "submittedVersion", "is_accepted": false, "is_published": false, "raw_source_name": null, "raw_type": "text"}, "cited_by_count": 0, "is_retracted": false}, {"id": "https://openalex.org/W4409352955", "doi": "https://doi.org/10.1101/2025.04.07.647599", "title": "Ontology-based Protein-Protein Interaction Explanation Using Large Language Models", "publication_year": 2025, "publication_date": "2025-04-11", "type": "preprint", "language": "en", "authorships": [{"author_position": "first", "author": {"id": "https://openalex.org/A5038162230", "display_name": "Nur Bengisu \u00c7am", "orcid": null, "observed_orcids": []}, "institutions": [{"id": "https://openalex.org/I4405392", "display_name": "Bo\u011fazi\u00e7i University", "ror": "https://ror.org/03z9tma90", "country_code": "TR", "type": "education", "lineage": ["https://openalex.org/I4405392"]}], "countries": ["TR"], "is_corresponding": false, "raw_author_name": "Nur Bengisu \u00c7am", "raw_affiliation_strings": ["Department of Computer Engineering, Bogazici University, Istanbul, T\u00fcrkiye"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Department of Computer Engineering, Bogazici University, Istanbul, T\u00fcrkiye", "institution_ids": ["https://openalex.org/I4405392"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5045250833", "display_name": "Hasin Rehana", "orcid": "https://orcid.org/0000-0003-2992-6547", "observed_orcids": ["https://orcid.org/0000-0003-2992-6547"]}, "institutions": [{"id": "https://openalex.org/I24571045", "display_name": "University of North Dakota", "ror": "https://ror.org/04a5szx83", "country_code": "US", "type": "education", "lineage": ["https://openalex.org/I24571045"]}], "countries": ["US"], "is_corresponding": false, "raw_author_name": "Hasin Rehana", "raw_affiliation_strings": ["Department of Computer Science, University of North Dakota, Grand Forks, USA"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Department of Computer Science, University of North Dakota, Grand Forks, USA", "institution_ids": ["https://openalex.org/I24571045"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5014529529", "display_name": "Jie Zheng", "orcid": "https://orcid.org/0000-0001-6774-9786", "observed_orcids": ["https://orcid.org/0000-0001-6774-9786"]}, "institutions": [{"id": "https://openalex.org/I27837315", "display_name": "University of Michigan", "ror": "https://ror.org/00jmfr291", "country_code": "US", "type": "education", "lineage": ["https://openalex.org/I27837315"]}], "countries": ["US"], "is_corresponding": false, "raw_author_name": "Jie Zheng", "raw_affiliation_strings": ["Unit for Laboratory Animal Medicine, University of Michigan, Ann Arbor, USA"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Unit for Laboratory Animal Medicine, University of Michigan, Ann Arbor, USA", "institution_ids": ["https://openalex.org/I27837315"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5043409464", "display_name": "Benu Bansal", "orcid": "https://orcid.org/0000-0002-2834-197X", "observed_orcids": ["https://orcid.org/0000-0002-2834-197X"]}, "institutions": [{"id": "https://openalex.org/I24571045", "display_name": "University of North Dakota", "ror": "https://ror.org/04a5szx83", "country_code": "US", "type": "education", "lineage": ["https://openalex.org/I24571045"]}], "countries": ["US"], "is_corresponding": false, "raw_author_name": "Benu Bansal", "raw_affiliation_strings": ["Department of Biomedical Engineering, University of North Dakota, Grand Forks, USA"], "raw_orcid": null, "affiliations": [{"raw_affiliation_string": "Department of Biomedical Engineering, University of North Dakota, Grand Forks, USA", "institution_ids": ["https://openalex.org/I24571045"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5024742201", "display_name": "Yongqun Oliver He", "orcid": "https://orcid.org/0000-0001-9189-9661", "observed_orcids": ["https://orcid.org/0000-0001-9189-9661"]}, "institutions": [{"id": "https://openalex.org/I27837315", "display_name": "University of Michigan", "ror": "https://ror.org/00jmfr291", "country_code": "US", "type": "education", "lineage": ["https://openalex.org/I27837315"]}], "countries": ["US"], "is_corresponding": false, "raw_author_name": "Yongqun He", "raw_affiliation_strings": ["Unit for Laboratory Animal Medicine, University of Michigan, Ann Arbor, USA"], "raw_orcid": "https://orcid.org/0000-0001-9189-9661", "affiliations": [{"raw_affiliation_string": "Unit for Laboratory Animal Medicine, University of Michigan, Ann Arbor, USA", "institution_ids": ["https://openalex.org/I27837315"]}]}, {"author_position": "middle", "author": {"id": "https://openalex.org/A5068376728", "display_name": "Junguk Hur", "orcid": "https://orcid.org/0000-0002-0736-2149", "observed_orcids": ["https://orcid.org/0000-0002-0736-2149"]}, "institutions": [{"id": "https://openalex.org/I24571045", "display_name": "University of North Dakota", "ror": "https://ror.org/04a5szx83", "country_code": "US", "type": "education", "lineage": ["https://openalex.org/I24571045"]}], "countries": ["US"], "is_corresponding": false, "raw_author_name": "Junguk Hur", "raw_affiliation_strings": ["Department of Biomedical Sciences, University of North Dakota, Grand Forks, USA"], "raw_orcid": "https://orcid.org/0000-0002-0736-2149", "affiliations": [{"raw_affiliation_string": "Department of Biomedical Sciences, University of North Dakota, Grand Forks, USA", "institution_ids": ["https://openalex.org/I24571045"]}]}, {"author_position": "last", "author": {"id": "https://openalex.org/A5030252794", "display_name": "Arzucan \u00d6zg\u00fcr", "orcid": "https://orcid.org/0000-0001-8376-1056", "observed_orcids": ["https://orcid.org/0000-0001-8376-1056"]}, "institutions": [{"id": "https://openalex.org/I4405392", "display_name": "Bo\u011fazi\u00e7i University", "ror": "https://ror.org/03z9tma90", "country_code": "TR", "type": "education", "lineage": ["https://openalex.org/I4405392"]}], "countries": ["TR"], "is_corresponding": false, "raw_author_name": "Arzucan \u00d6zg\u00fcr", "raw_affiliation_strings": ["Department of Computer Engineering, Bogazici University, Istanbul, T\u00fcrkiye"], "raw_orcid": "https://orcid.org/0000-0001-8376-1056", "affiliations": [{"raw_affiliation_string": "Department of Computer Engineering, Bogazici University, Istanbul, T\u00fcrkiye", "institution_ids": ["https://openalex.org/I4405392"]}]}], "abstract_inverted_index": {"Abstract": [0], "Protein-protein": [1], "interactions": [2, 15, 48], "(PPIs)": [3], "play": [4], "a": [5, 41, 82, 94, 138], "crucial": [6], "role": [7], "in": [8, 74, 128], "various": [9], "biological": [10], "processes,": [11], "and": [12, 24, 62, 143, 164], "understanding": [13], "these": [14], "is": [16, 101, 146], "essential": [17], "for": [18, 66, 150], "advancing": [19], "biomedical": [20, 154], "research.": [21], "Automated": [22], "extraction": [23], "analysis": [25, 172], "of": [26, 84, 111, 140, 161], "PPI": [27, 171], "information": [28], "from": [29, 173], "the": [30, 67, 75, 99, 108, 119, 125, 129, 159, 174], "rapidly": [31], "growing": [32], "scientific": [33, 175], "literature": [34], "remains": [35], "an": [36, 79, 141, 144, 147], "important": [37], "challenge.": [38], "We": [39, 54], "present": [40], "novel": [42], "ontology-based": [43], "approach": [44], "to": [45, 71, 93, 124, 168], "analyze": [46], "protein-protein": [47], "using": [49], "Large": [50], "Language": [51], "Models": [52], "(LLMs).": [53], "applied": [55], "different": [56], "learning": [57, 61], "strategies,": [58], "namely": [59], "in-context": [60], "parameter-efficient": [63, 90], "instruction": [64], "fine-tuning": [65, 91], "Llama-2": [68, 120], "chat": [69], "models,": [70], "identify": [72], "keywords": [73, 116], "text": [76], "that": [77, 89, 137], "indicate": [78], "interaction": [80], "between": [81, 153], "pair": [83], "proteins.": [85], "Our": [86, 134], "results": [87], "show": [88], "leads": [92], "performance": [95, 110], "gain": [96], "even": [97], "when": [98], "domain": [100], "new.": [102], "The": [103, 115], "smaller": [104], "fine-tuned": [105], "models": [106, 121, 167], "outperformed": [107], "zero-shot": [109], "much": [112], "larger": [113], "models.": [114], "identified": [117], "by": [118], "were": [122], "mapped": [123], "ontology": [126, 145], "terms": [127], "Interaction": [130], "Network": [131], "Ontology": [132], "(INO).": [133], "study": [135], "suggests": [136], "pipeline": [139], "LLM": [142], "effective": [148], "strategy": [149], "explaining": [151], "relations": [152], "entities.": [155], "This": [156], "work": [157], "demonstrates": [158], "potential": [160], "leveraging": [162], "ontologies": [163], "advanced": [165], "language": [166], "advance": [169], "automated": [170], "literature.": [176]}, "primary_topic": {"id": "https://openalex.org/T11710", "display_name": "Biomedical Text Mining and Ontologies", "score": 0.9976000189781189, "subfield": {"id": "https://openalex.org/subfields/1312", "display_name": "Molecular Biology"}, "field": {"id": "https://openalex.org/fields/13", "display_name": "Biochemistry, Genetics and Molecular Biology"}, "domain": {"id": "https://openalex.org/domains/1", "display_name": "Life Sciences"}}, "topics": [{"id": "https://openalex.org/T11710", "display_name": "Biomedical Text Mining and Ontologies", "score": 0.9976000189781189, "subfield": {"id": "https://openalex.org/subfields/1312", "display_name": "Molecular Biology"}, "field": {"id": "https://openalex.org/fields/13", "display_name": "Biochemistry, Genetics and Molecular Biology"}, "domain": {"id": "https://openalex.org/domains/1", "display_name": "Life Sciences"}}, {"id": "https://openalex.org/T10887", "display_name": "Bioinformatics and Genomic Networks", "score": 0.00139999995008111, "subfield": {"id": "https://openalex.org/subfields/1312", "display_name": "Molecular Biology"}, "field": {"id": "https://openalex.org/fields/13", "display_name": "Biochemistry, Genetics and Molecular Biology"}, "domain": {"id": "https://openalex.org/domains/1", "display_name": "Life Sciences"}}, {"id": "https://openalex.org/T10215", "display_name": "Semantic Web and Ontologies", "score": 9.999999747378752e-05, "subfield": {"id": "https://openalex.org/subfields/1702", "display_name": "Artificial Intelligence"}, "field": {"id": "https://openalex.org/fields/17", "display_name": "Computer Science"}, "domain": {"id": "https://openalex.org/domains/3", "display_name": "Physical Sciences"}}], "primary_location": {"id": "doi:10.1101/2025.04.07.647599", "is_oa": true, "landing_page_url": "https://doi.org/10.1101/2025.04.07.647599", "pdf_url": null, "source": {"id": "https://openalex.org/S4306402567", "display_name": "bioRxiv (Cold Spring Harbor Laboratory)", "issn_l": null, "issn": null, "is_oa": false, "is_in_doaj": false, "is_core": false, "listed_in": [], "host_organization": "https://openalex.org/I2750212522", "host_organization_name": "Cold Spring Harbor Laboratory", "host_organization_lineage": ["https://openalex.org/I2750212522"], "host_organization_lineage_names": [], "type": "repository"}, "license": "cc-by", "license_id": "https://openalex.org/licenses/cc-by", "version": "acceptedVersion", "is_accepted": true, "is_published": false, "raw_source_name": "bioRxiv", "raw_type": "posted-content"}, "cited_by_count": 0, "is_retracted": false}], "group_by": []}