{"id":"https://openalex.org/W4416036388","doi":"https://doi.org/10.18653/v1/2025.emnlp-main.581","title":"Improving Instruct Models for Free: A Study on Partial Adaptation","display_name":"Improving Instruct Models for Free: A Study on Partial Adaptation","publication_year":2025,"publication_date":"2025-01-01","ids":{"openalex":"https://openalex.org/W4416036388","doi":"https://doi.org/10.18653/v1/2025.emnlp-main.581"},"language":null,"primary_location":{"id":"doi:10.18653/v1/2025.emnlp-main.581","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.emnlp-main.581","pdf_url":"https://aclanthology.org/2025.emnlp-main.581.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":"Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing","raw_type":"proceedings-article"},"type":"article","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://aclanthology.org/2025.emnlp-main.581.pdf","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5088262368","display_name":"Ozan \u0130rsoy","orcid":"https://orcid.org/0000-0002-7123-8361"},"institutions":[],"countries":[],"is_corresponding":true,"raw_author_name":"Ozan Irsoy","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5088553265","display_name":"Pengxiang Cheng","orcid":"https://orcid.org/0000-0001-5997-705X"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Pengxiang Cheng","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5033706964","display_name":"Jennifer L. Chen","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jennifer L Chen","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5059811412","display_name":"Daniel Preo\u0163iuc-Pietro","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Daniel Preotiuc-Pietro","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5052879411","display_name":"Shengbai Zhang","orcid":"https://orcid.org/0000-0003-0833-5860"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Shiyue Zhang","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5033084003","display_name":"Duccio Pappadopulo","orcid":"https://orcid.org/0000-0003-3037-759X"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Duccio Pappadopulo","raw_affiliation_strings":[],"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":6,"corresponding_author_ids":["https://openalex.org/A5088262368"],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":true,"cited_by_count":0,"citation_normalized_percentile":{"value":0.18434658,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"11518","last_page":"11532"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11902","display_name":"Intelligent Tutoring Systems and Adaptive Learning","score":0.07670000195503235,"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/T11902","display_name":"Intelligent Tutoring Systems and Adaptive Learning","score":0.07670000195503235,"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/T11574","display_name":"Artificial Intelligence in Games","score":0.04729999974370003,"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/T10826","display_name":"Behavioral and Psychological Studies","score":0.04050000011920929,"subfield":{"id":"https://openalex.org/subfields/3204","display_name":"Developmental and Educational Psychology"},"field":{"id":"https://openalex.org/fields/32","display_name":"Psychology"},"domain":{"id":"https://openalex.org/domains/2","display_name":"Social Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/adaptation","display_name":"Adaptation (eye)","score":0.3668999969959259},{"id":"https://openalex.org/keywords/set","display_name":"Set (abstract data type)","score":0.25429999828338623},{"id":"https://openalex.org/keywords/context","display_name":"Context (archaeology)","score":0.2526000142097473},{"id":"https://openalex.org/keywords/perspective","display_name":"Perspective (graphical)","score":0.2524000108242035},{"id":"https://openalex.org/keywords/process","display_name":"Process (computing)","score":0.25049999356269836}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5440000295639038},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.4090999960899353},{"id":"https://openalex.org/C139807058","wikidata":"https://www.wikidata.org/wiki/Q352374","display_name":"Adaptation (eye)","level":2,"score":0.3668999969959259},{"id":"https://openalex.org/C177264268","wikidata":"https://www.wikidata.org/wiki/Q1514741","display_name":"Set (abstract data type)","level":2,"score":0.25429999828338623},{"id":"https://openalex.org/C2779343474","wikidata":"https://www.wikidata.org/wiki/Q3109175","display_name":"Context (archaeology)","level":2,"score":0.2526000142097473},{"id":"https://openalex.org/C12713177","wikidata":"https://www.wikidata.org/wiki/Q1900281","display_name":"Perspective (graphical)","level":2,"score":0.2524000108242035},{"id":"https://openalex.org/C98045186","wikidata":"https://www.wikidata.org/wiki/Q205663","display_name":"Process (computing)","level":2,"score":0.25049999356269836},{"id":"https://openalex.org/C165064840","wikidata":"https://www.wikidata.org/wiki/Q1321061","display_name":"Matching (statistics)","level":2,"score":0.23520000278949738},{"id":"https://openalex.org/C2779530757","wikidata":"https://www.wikidata.org/wiki/Q1207505","display_name":"Quality (philosophy)","level":2,"score":0.22419999539852142},{"id":"https://openalex.org/C61797465","wikidata":"https://www.wikidata.org/wiki/Q1188986","display_name":"Term (time)","level":2,"score":0.21770000457763672}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.18653/v1/2025.emnlp-main.581","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.emnlp-main.581","pdf_url":"https://aclanthology.org/2025.emnlp-main.581.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":"Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.18653/v1/2025.emnlp-main.581","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.emnlp-main.581","pdf_url":"https://aclanthology.org/2025.emnlp-main.581.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":"Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4416036388.pdf","grobid_xml":"https://content.openalex.org/works/W4416036388.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Instruct":[0],"models,":[1],"obtained":[2],"from":[3,35,286],"various":[4],"instruction":[5,24,27,125,143],"tuning":[6,28],"or":[7,37,47],"post-training":[8],"steps,":[9],"are":[10],"commonly":[11],"deemed":[12],"superior":[13],"and":[14,69,90,142],"more":[15],"usable":[16],"than":[17],"their":[18],"base":[19,68],"counterpart.While":[20],"the":[21,33,41,64,75,80,94,118,136],"model":[22,42,88,91,288],"gains":[23],"following":[25,126,144],"ability,":[26],"may":[29,39],"lead":[30,52],"to":[31,43,53],"forgetting":[32],"knowledge":[34],"pre-training":[36],"it":[38],"encourage":[40],"become":[44],"overly":[45],"conversational":[46],"verbose.This,":[48],"in":[49,99,150],"turn,":[50],"can":[51],"degradation":[54],"of":[55,77,96,111,120,124],"in-context":[56,105,140],"few-shot":[57,104],"learning":[58,106,141],"performance.In":[59],"this":[60],"work,":[61],"we":[62],"study":[63,132],"performance":[65],"trajectory":[66],"between":[67,139],"instruct":[70,287],"models":[71],"by":[72,130],"scaling":[73],"down":[74],"strength":[76,95],"instruction-tuning":[78,97],"via":[79],"partial":[81],"adaption":[82],"method.We":[83],"show":[84],"that,":[85],"across":[86],"several":[87],"families":[89],"sizes,":[92],"reducing":[93],"results":[98],"material":[100],"improvement":[101],"on":[102,135],"a":[103,109],"benchmark":[107],"covering":[108],"variety":[110],"classic":[112],"natural":[113],"language":[114],"tasks.This":[115],"comes":[116],"at":[117,155,160],"cost":[119],"losing":[121],"some":[122],"degree":[123],"ability":[127],"as":[128],"measured":[129],"AlpacaEval.Our":[131],"shines":[133],"light":[134],"potential":[137],"trade-off":[138],"abilities":[145],"that":[146],"is":[147],"worth":[148],"considering":[149],"practice.*":[151],"Work":[152],"done":[153],"while":[154],"Bloomberg.":[156],"Author":[157],"ordering":[158],"chosen":[159],"random.0.0":[161],"0.2":[162,176,190,204,220,239,255,273],"0.4":[163,177,191,205,221,240,256,274],"0.6":[164,178,192,206,222,241,257,275],"0.8":[165,179,193,207,223,242,258,276],"1.0":[166,180,194,208,224,243,259,277],"5":[167,169,181,183,195,197,209,211,225,227,244,246,260,262,278,280],"0":[168,182,196,210,226,245,261,279],"10":[170,184,198,212,228,247,263,281],"Llama-2":[171,173],"70B":[172,186,200,218],"7B":[174,230,233,265,268],"0.0":[175,189,203,219,238,254,272],"Llama-3":[185,187],"8B":[188,202],"Llama-3.1":[199,201],"Llama-3.2":[213,215],"1B":[214],"3B":[216],"Llama-3.3":[217],"Mistral":[229,232,235],"v0.3":[231],"v0.1":[234,250,253],"Nemo":[236],"12B":[237],"Mixtral":[248,251],"8x22B":[249],"8x7B":[252],"OLMo":[264,266,269],"2":[267,270],"13B":[271],"Gemma-2":[282],"9B":[283],"%":[284],"change":[285],"(M":[289],"1":[290],")":[291]},"counts_by_year":[],"updated_date":"2026-03-11T06:11:40.159057","created_date":"2025-11-08T00:00:00"}