{"id":"https://openalex.org/W4416034934","doi":"https://doi.org/10.18653/v1/2025.findings-emnlp.642","title":"ReAlign: Structured Revision for Small Language Model Alignment","display_name":"ReAlign: Structured Revision for Small Language Model Alignment","publication_year":2025,"publication_date":"2025-01-01","ids":{"openalex":"https://openalex.org/W4416034934","doi":"https://doi.org/10.18653/v1/2025.findings-emnlp.642"},"language":null,"primary_location":{"id":"doi:10.18653/v1/2025.findings-emnlp.642","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-emnlp.642","pdf_url":"https://aclanthology.org/2025.findings-emnlp.642.pdf","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Findings of the Association for Computational Linguistics: EMNLP 2025","raw_type":"proceedings-article"},"type":"conference-paper","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://aclanthology.org/2025.findings-emnlp.642.pdf","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5043323192","display_name":"Ruijun Chen","orcid":"https://orcid.org/0000-0001-5880-7446"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ruijun Chen","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Jiajian Guo","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jiajian Guo","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5012382750","display_name":"Hongzhan Chen","orcid":"https://orcid.org/0009-0009-7622-2244"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Hongzhan Chen","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5101262140","display_name":"Fanqi Wan","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Fanqi Wan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5055007304","display_name":"Qifan Wang","orcid":"https://orcid.org/0000-0002-5304-7975"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Qifan Wang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5040062188","display_name":"Xiaojun Quan","orcid":"https://orcid.org/0000-0002-8385-1083"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Xiaojun Quan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":0,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":true,"cited_by_count":0,"citation_normalized_percentile":{"value":0.3004688,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"12005","last_page":"12020"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10028","display_name":"Topic Modeling","score":0.3544999957084656,"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":0.3544999957084656,"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":0.2824999988079071,"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/T10201","display_name":"Speech Recognition and Synthesis","score":0.05339999869465828,"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"}}],"keywords":[{"id":"https://openalex.org/keywords/natural-language","display_name":"Natural language","score":0.32739999890327454},{"id":"https://openalex.org/keywords/language-model","display_name":"Language model","score":0.3151000142097473},{"id":"https://openalex.org/keywords/feature","display_name":"Feature (linguistics)","score":0.3102000057697296},{"id":"https://openalex.org/keywords/translation","display_name":"Translation (biology)","score":0.26750001311302185},{"id":"https://openalex.org/keywords/set","display_name":"Set (abstract data type)","score":0.2667999863624573}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6549999713897705},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5012999773025513},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.47350001335144043},{"id":"https://openalex.org/C195324797","wikidata":"https://www.wikidata.org/wiki/Q33742","display_name":"Natural language","level":2,"score":0.32739999890327454},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.3151000142097473},{"id":"https://openalex.org/C2776401178","wikidata":"https://www.wikidata.org/wiki/Q12050496","display_name":"Feature (linguistics)","level":2,"score":0.3102000057697296},{"id":"https://openalex.org/C41895202","wikidata":"https://www.wikidata.org/wiki/Q8162","display_name":"Linguistics","level":1,"score":0.3037000000476837},{"id":"https://openalex.org/C149364088","wikidata":"https://www.wikidata.org/wiki/Q185917","display_name":"Translation (biology)","level":4,"score":0.26750001311302185},{"id":"https://openalex.org/C177264268","wikidata":"https://www.wikidata.org/wiki/Q1514741","display_name":"Set (abstract data type)","level":2,"score":0.2667999863624573},{"id":"https://openalex.org/C67186912","wikidata":"https://www.wikidata.org/wiki/Q367664","display_name":"Data modeling","level":2,"score":0.2660999894142151},{"id":"https://openalex.org/C184337299","wikidata":"https://www.wikidata.org/wiki/Q1437428","display_name":"Semantics (computer science)","level":2,"score":0.25369998812675476}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.18653/v1/2025.findings-emnlp.642","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-emnlp.642","pdf_url":"https://aclanthology.org/2025.findings-emnlp.642.pdf","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Findings of the Association for Computational Linguistics: EMNLP 2025","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.18653/v1/2025.findings-emnlp.642","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-emnlp.642","pdf_url":"https://aclanthology.org/2025.findings-emnlp.642.pdf","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Findings of the Association for Computational Linguistics: EMNLP 2025","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[{"id":"https://openalex.org/G6792809764","display_name":null,"funder_award_id":"62176270","funder_id":"https://openalex.org/F4320321001","funder_display_name":"National Natural Science Foundation of China"},{"id":"https://openalex.org/G7447199816","display_name":null,"funder_award_id":"2023A1515012832","funder_id":"https://openalex.org/F4320337111","funder_display_name":"Basic and Applied Basic Research Foundation of Guangdong Province"}],"funders":[{"id":"https://openalex.org/F4320321001","display_name":"National Natural Science Foundation of China","ror":"https://ror.org/01h0zpd94"},{"id":"https://openalex.org/F4320337111","display_name":"Basic and Applied Basic Research Foundation of Guangdong Province","ror":null}],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4416034934.pdf","grobid_xml":"https://content.openalex.org/works/W4416034934.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Aligning":[0],"small":[1,128],"language":[2,129],"models":[3,31,130],"with":[4,50],"human":[5,33],"preferences":[6],"remains":[7],"a":[8,40,57,85,99],"challenging":[9],"problem:":[10],"weak":[11],"policies":[12],"often":[13],"struggle":[14],"to":[15,64],"produce":[16],"informative":[17,108],"on-policy":[18,48,91],"samples":[19],"and":[20,76,110,118,131],"exhibit":[21],"unstable":[22],"gradients":[23],"when":[24],"trained":[25,63],"on":[26,72,116],"off-policy":[27],"signals":[28],"from":[29],"stronger":[30],"or":[32],"annotators.In":[34],"this":[35],"work,":[36],"we":[37],"introduce":[38],"ReAlign,":[39,56],"training":[41],"framework":[42],"that":[43,88,121],"combines":[44],"the":[45,51,74,77],"stability":[46],"of":[47,53],"learning":[49],"guidance":[52],"reviser-assisted":[54],"supervision.In":[55],"lightweight":[58],"external":[59],"reviser":[60],"is":[61,81],"first":[62],"improve":[65],"policy-generated":[66],"responses":[67],"using":[68,84],"preference-based":[69],"feedback,":[70],"conditioned":[71],"both":[73],"prompt":[75],"initial":[78],"output.The":[79],"policy":[80],"then":[82],"optimized":[83],"hybrid":[86],"approach":[87],"leverages":[89],"standard":[90],"preference":[92,134],"pairs":[93,96,104],"alongside":[94],"reviser-enhanced":[95],"framed":[97],"as":[98],"structured":[100],"revision":[101],"task.These":[102],"enhanced":[103],"provide":[105],"richer,":[106],"more":[107,112],"supervision,":[109],"facilitate":[111],"effective":[113],"optimization.Extensive":[114],"experiments":[115],"AlpacaEval-2":[117],"Arena-Hard":[119],"demonstrate":[120],"ReAlign":[122],"consistently":[123],"improves":[124],"alignment":[125],"performance":[126],"for":[127],"outperforms":[132],"strong":[133],"optimization":[135],"baselines.":[136]},"counts_by_year":[],"updated_date":"2026-07-29T14:22:42.915294","created_date":"2025-11-08T00:00:00"}