{"id":"https://openalex.org/W4389520751","doi":"https://doi.org/10.18653/v1/2023.emnlp-main.559","title":"Revisiting Instruction Fine-tuned Model Evaluation to Guide Industrial Applications","display_name":"Revisiting Instruction Fine-tuned Model Evaluation to Guide Industrial Applications","publication_year":2023,"publication_date":"2023-01-01","ids":{"openalex":"https://openalex.org/W4389520751","doi":"https://doi.org/10.18653/v1/2023.emnlp-main.559"},"language":"en","primary_location":{"id":"doi:10.18653/v1/2023.emnlp-main.559","doi":"https://doi.org/10.18653/v1/2023.emnlp-main.559","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2023.emnlp-main.559","pdf_url":"https://aclanthology.org/2023.emnlp-main.559.pdf","source":{"id":"https://openalex.org/S4363608991","display_name":"Conference on Empirical Methods in Natural Language Processing (EMNLP)","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":"conference"},"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 2023 Conference on Empirical Methods in Natural Language Processing","raw_type":"proceedings-article"},"volume":null,"type":"conference-paper","indexed_in":["arxiv","crossref","datacite"],"open_access":{"is_oa":true,"oa_status":"hybrid","oa_url":"https://aclanthology.org/2023.emnlp-main.559.pdf","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5093123135","display_name":"Manuel Faysse","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I277688954","display_name":"Universit\u00e9 Paris-Saclay","ror":"https://ror.org/03xjwb503","country_code":"FR","type":"education","lineage":["https://openalex.org/I277688954"]},{"id":"https://openalex.org/I4210107720","display_name":"CentraleSup\u00e9lec","ror":"https://ror.org/019tcpt25","country_code":"FR","type":"facility","lineage":["https://openalex.org/I277688954","https://openalex.org/I4210107720"]},{"id":"https://openalex.org/I4387152683","display_name":"Math\u00e9matiques et Informatique pour la Complexit\u00e9 et les Syst\u00e8mes","ror":"https://ror.org/03q29s414","country_code":"FR","type":"facility","lineage":["https://openalex.org/I277688954","https://openalex.org/I4210107720","https://openalex.org/I4387152683"]}],"countries":["FR"],"is_corresponding":false,"raw_author_name":"Manuel Faysse","raw_affiliation_strings":["Illuin Technology , Paris , France","MICS , CentraleSup\u00e9lec , Universit\u00e9 Paris-Saclay , France","MICS - Math\u00e9matiques et Informatique pour la Complexit\u00e9 et les Syst\u00e8mes (Plateau de Moulon 3 rue Joliot Curie 91192 GIF SUR YVETTE CEDEX - France)"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Illuin Technology , Paris , France","institution_ids":[]},{"raw_affiliation_string":"MICS , CentraleSup\u00e9lec , Universit\u00e9 Paris-Saclay , France","institution_ids":["https://openalex.org/I277688954","https://openalex.org/I4210107720"]},{"raw_affiliation_string":"MICS - Math\u00e9matiques et Informatique pour la Complexit\u00e9 et les Syst\u00e8mes (Plateau de Moulon 3 rue Joliot Curie 91192 GIF SUR YVETTE CEDEX - France)","institution_ids":["https://openalex.org/I4387152683"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5079512127","display_name":"Gautier Viaud","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I4387152683","display_name":"Math\u00e9matiques et Informatique pour la Complexit\u00e9 et les Syst\u00e8mes","ror":"https://ror.org/03q29s414","country_code":"FR","type":"facility","lineage":["https://openalex.org/I277688954","https://openalex.org/I4210107720","https://openalex.org/I4387152683"]}],"countries":["FR"],"is_corresponding":false,"raw_author_name":"Gautier Viaud","raw_affiliation_strings":["Illuin Technology , Paris , France","MICS - Math\u00e9matiques et Informatique pour la Complexit\u00e9 et les Syst\u00e8mes (Plateau de Moulon 3 rue Joliot Curie 91192 GIF SUR YVETTE CEDEX - France)"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Illuin Technology , Paris , France","institution_ids":[]},{"raw_affiliation_string":"MICS - Math\u00e9matiques et Informatique pour la Complexit\u00e9 et les Syst\u00e8mes (Plateau de Moulon 3 rue Joliot Curie 91192 GIF SUR YVETTE CEDEX - France)","institution_ids":["https://openalex.org/I4387152683"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5070164335","display_name":"C\u00e9line Hudelot","orcid":"https://orcid.org/0000-0003-3849-4133","observed_orcids":["https://orcid.org/0000-0003-3849-4133"]},"institutions":[{"id":"https://openalex.org/I277688954","display_name":"Universit\u00e9 Paris-Saclay","ror":"https://ror.org/03xjwb503","country_code":"FR","type":"education","lineage":["https://openalex.org/I277688954"]},{"id":"https://openalex.org/I4210107720","display_name":"CentraleSup\u00e9lec","ror":"https://ror.org/019tcpt25","country_code":"FR","type":"facility","lineage":["https://openalex.org/I277688954","https://openalex.org/I4210107720"]},{"id":"https://openalex.org/I4387152683","display_name":"Math\u00e9matiques et Informatique pour la Complexit\u00e9 et les Syst\u00e8mes","ror":"https://ror.org/03q29s414","country_code":"FR","type":"facility","lineage":["https://openalex.org/I277688954","https://openalex.org/I4210107720","https://openalex.org/I4387152683"]}],"countries":["FR"],"is_corresponding":false,"raw_author_name":"C\u00e9line Hudelot","raw_affiliation_strings":["MICS , CentraleSup\u00e9lec , Universit\u00e9 Paris-Saclay , France","MICS - Math\u00e9matiques et Informatique pour la Complexit\u00e9 et les Syst\u00e8mes (Plateau de Moulon 3 rue Joliot Curie 91192 GIF SUR YVETTE CEDEX - France)"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"MICS , CentraleSup\u00e9lec , Universit\u00e9 Paris-Saclay , France","institution_ids":["https://openalex.org/I277688954","https://openalex.org/I4210107720"]},{"raw_affiliation_string":"MICS - Math\u00e9matiques et Informatique pour la Complexit\u00e9 et les Syst\u00e8mes (Plateau de Moulon 3 rue Joliot Curie 91192 GIF SUR YVETTE CEDEX - France)","institution_ids":["https://openalex.org/I4387152683"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5103969906","display_name":"Pierre Colombo","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I277688954","display_name":"Universit\u00e9 Paris-Saclay","ror":"https://ror.org/03xjwb503","country_code":"FR","type":"education","lineage":["https://openalex.org/I277688954"]},{"id":"https://openalex.org/I4210107720","display_name":"CentraleSup\u00e9lec","ror":"https://ror.org/019tcpt25","country_code":"FR","type":"facility","lineage":["https://openalex.org/I277688954","https://openalex.org/I4210107720"]},{"id":"https://openalex.org/I4387152683","display_name":"Math\u00e9matiques et Informatique pour la Complexit\u00e9 et les Syst\u00e8mes","ror":"https://ror.org/03q29s414","country_code":"FR","type":"facility","lineage":["https://openalex.org/I277688954","https://openalex.org/I4210107720","https://openalex.org/I4387152683"]}],"countries":["FR"],"is_corresponding":false,"raw_author_name":"Pierre Colombo","raw_affiliation_strings":["Equall , Paris , France","MICS , CentraleSup\u00e9lec , Universit\u00e9 Paris-Saclay , France","Equall (France)","MICS - Math\u00e9matiques et Informatique pour la Complexit\u00e9 et les Syst\u00e8mes (Plateau de Moulon 3 rue Joliot Curie 91192 GIF SUR YVETTE CEDEX - France)"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Equall , Paris , France","institution_ids":[]},{"raw_affiliation_string":"MICS , CentraleSup\u00e9lec , Universit\u00e9 Paris-Saclay , France","institution_ids":["https://openalex.org/I277688954","https://openalex.org/I4210107720"]},{"raw_affiliation_string":"Equall (France)","institution_ids":[]},{"raw_affiliation_string":"MICS - Math\u00e9matiques et Informatique pour la Complexit\u00e9 et les Syst\u00e8mes (Plateau de Moulon 3 rue Joliot Curie 91192 GIF SUR YVETTE CEDEX - France)","institution_ids":["https://openalex.org/I4387152683"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":3,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":0.5289,"has_fulltext":true,"cited_by_count":3,"citation_normalized_percentile":{"value":0.66090449,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":89,"max":96},"biblio":{"volume":null,"issue":null,"first_page":"9033","last_page":"9048"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10028","display_name":"Topic Modeling","score":0.9009000062942505,"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":0.9009000062942505,"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":0.05620000138878822,"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":0.015799999237060547,"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/instruction-tuning","display_name":"instruction tuning","score":0.996999979019165},{"id":"https://openalex.org/keywords/large-language-models","display_name":"large language models","score":0.9679999947547913},{"id":"https://openalex.org/keywords/zero-shot-learning","display_name":"zero-shot learning","score":0.6050000190734863},{"id":"https://openalex.org/keywords/task-specialization","display_name":"task specialization","score":0.6100000143051147},{"id":"https://openalex.org/keywords/industrial-applications","display_name":"industrial applications","score":0.47699999809265137},{"id":"https://openalex.org/keywords/model-deployment","display_name":"model deployment","score":0.7820000052452087},{"id":"https://openalex.org/keywords/trade-offs","display_name":"trade-offs","score":0.5569999814033508}],"concepts":[{"id":"https://openalex.org/C153083717","wikidata":"https://www.wikidata.org/wiki/Q6535263","display_name":"Leverage (statistics)","level":2,"score":0.8411526679992676},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.670839786529541},{"id":"https://openalex.org/C105339364","wikidata":"https://www.wikidata.org/wiki/Q2297740","display_name":"Software deployment","level":2,"score":0.6642821431159973},{"id":"https://openalex.org/C176217482","wikidata":"https://www.wikidata.org/wiki/Q860554","display_name":"Metric (unit)","level":2,"score":0.5444759130477905},{"id":"https://openalex.org/C2780451532","wikidata":"https://www.wikidata.org/wiki/Q759676","display_name":"Task (project management)","level":2,"score":0.534047544002533},{"id":"https://openalex.org/C115903868","wikidata":"https://www.wikidata.org/wiki/Q80993","display_name":"Software engineering","level":1,"score":0.3033260107040405},{"id":"https://openalex.org/C201995342","wikidata":"https://www.wikidata.org/wiki/Q682496","display_name":"Systems engineering","level":1,"score":0.2785240411758423},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.22255656123161316},{"id":"https://openalex.org/C127413603","wikidata":"https://www.wikidata.org/wiki/Q11023","display_name":"Engineering","level":0,"score":0.1901177167892456},{"id":"https://openalex.org/C21547014","wikidata":"https://www.wikidata.org/wiki/Q1423657","display_name":"Operations management","level":1,"score":0.09998595714569092}],"mesh":[],"locations_count":5,"locations":[{"id":"doi:10.18653/v1/2023.emnlp-main.559","doi":"https://doi.org/10.18653/v1/2023.emnlp-main.559","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2023.emnlp-main.559","pdf_url":"https://aclanthology.org/2023.emnlp-main.559.pdf","source":{"id":"https://openalex.org/S4363608991","display_name":"Conference on Empirical Methods in Natural Language Processing (EMNLP)","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":"conference"},"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 2023 Conference on Empirical Methods in Natural Language Processing","raw_type":"proceedings-article"},{"id":"pmh:oai:HAL:hal-04575351v1","doi":null,"is_oa":true,"landing_page_url":"https://hal.science/hal-04575351","pdf_url":"http://arxiv.org/pdf/2310.14103","source":{"id":"https://openalex.org/S4306402512","display_name":"HAL","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I4405252976","host_organization_name":"Centre pour la Communication Scientifique Directe","host_organization_lineage":["https://openalex.org/I4405252976"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":"Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing","raw_type":"info:eu-repo/semantics/conferenceObject"},{"id":"pmh:oai:arXiv.org:2310.14103","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2310.14103","pdf_url":"https://arxiv.org/pdf/2310.14103","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"},{"id":"pmh:oai:HAL:hal-04574148v1","doi":null,"is_oa":false,"landing_page_url":"https://hal.science/hal-04574148","pdf_url":null,"source":{"id":"https://openalex.org/S4306402512","display_name":"HAL","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I4405252976","host_organization_name":"Centre pour la Communication Scientifique Directe","host_organization_lineage":["https://openalex.org/I4405252976"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":"Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing, Dec 2023, Singapore, France. pp.9033-9048, ⟨10.18653/v1/2023.emnlp-main.559⟩","raw_type":"info:eu-repo/semantics/conferenceObject"},{"id":"doi:10.48550/arxiv.2310.14103","doi":"https://doi.org/10.48550/arxiv.2310.14103","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2310.14103","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":null,"license_id":null,"version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"article-journal"}],"best_oa_location":{"id":"doi:10.18653/v1/2023.emnlp-main.559","doi":"https://doi.org/10.18653/v1/2023.emnlp-main.559","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2023.emnlp-main.559","pdf_url":"https://aclanthology.org/2023.emnlp-main.559.pdf","source":{"id":"https://openalex.org/S4363608991","display_name":"Conference on Empirical Methods in Natural Language Processing (EMNLP)","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":"conference"},"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 2023 Conference on Empirical Methods in Natural Language Processing","raw_type":"proceedings-article"},"sustainable_development_goals":[],"study_designs":[],"awards":[],"funders":[{"id":"https://openalex.org/F4320321663","display_name":"Association Nationale de la Recherche et de la Technologie","ror":"https://ror.org/00ht2ab73"},{"id":"https://openalex.org/F4320326256","display_name":"Grand \u00c9quipement National De Calcul Intensif","ror":"https://ror.org/0010d1q40"}],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4389520751.pdf","grobid_xml":"https://content.openalex.org/works/W4389520751.grobid-xml"},"referenced_works_count":41,"referenced_works":["https://openalex.org/W2101105183","https://openalex.org/W2152180407","https://openalex.org/W2154652894","https://openalex.org/W2607892599","https://openalex.org/W2888482885","https://openalex.org/W2923014074","https://openalex.org/W2936695845","https://openalex.org/W2946609015","https://openalex.org/W2952087486","https://openalex.org/W2963748441","https://openalex.org/W2970641574","https://openalex.org/W3045321166","https://openalex.org/W3093429745","https://openalex.org/W3121904249","https://openalex.org/W3168867926","https://openalex.org/W3196731672","https://openalex.org/W3198002980","https://openalex.org/W3211686893","https://openalex.org/W4221167718","https://openalex.org/W4225576933","https://openalex.org/W4226278401","https://openalex.org/W4293333030","https://openalex.org/W4322718191","https://openalex.org/W4322760121","https://openalex.org/W4360836968","https://openalex.org/W4362655426","https://openalex.org/W4362707064","https://openalex.org/W4366341216","https://openalex.org/W4367000491","https://openalex.org/W4378189609","https://openalex.org/W4379468930","https://openalex.org/W4384918448","https://openalex.org/W4385571124","https://openalex.org/W4385572634","https://openalex.org/W4385573073","https://openalex.org/W4387635896","https://openalex.org/W4389518754","https://openalex.org/W4389519254","https://openalex.org/W4404792830","https://openalex.org/W7133189643","https://openalex.org/W7133220558"],"related_works":["https://openalex.org/W7115002957","https://openalex.org/W4381573334","https://openalex.org/W7161180357","https://openalex.org/W4388964644","https://openalex.org/W4312205996","https://openalex.org/W7204919635","https://openalex.org/W4389524167","https://openalex.org/W4416296080","https://openalex.org/W4386272802","https://openalex.org/W4388718089"],"abstract_inverted_index":{"Instruction":[0],"Fine-Tuning":[1],"(IFT)":[2],"is":[3],"a":[4],"powerful":[5],"paradigm":[6],"that":[7,49],"strengthens":[8],"the":[9,47],"zero-shot":[10],"capabilities":[11],"of":[12,43],"Large":[13],"Language":[14],"Models":[15],"(LLMs),":[16],"but":[17],"in":[18,51],"doing":[19],"so":[20],"induces":[21],"new":[22],"evaluation":[23],"metric":[24],"requirements.We":[25],"show":[26],"LLM-based":[27],"metrics":[28],"to":[29,33,39],"be":[30],"well":[31],"adapted":[32],"these":[34],"requirements,":[35],"and":[36],"leverage":[37],"them":[38],"conduct":[40],"an":[41],"investigation":[42],"taskspecialization":[44],"strategies,":[45],"quantifying":[46],"tradeoffs":[48],"emerge":[50],"practical":[52],"industrial":[53],"settings.Our":[54],"findings":[55],"offer":[56],"practitioners":[57],"actionable":[58],"insights":[59],"for":[60],"real-world":[61],"IFT":[62],"model":[63],"deployment.":[64]},"counts_by_year":[{"year":2026,"cited_by_count":1},{"year":2025,"cited_by_count":1},{"year":2024,"cited_by_count":1}],"updated_date":"2026-11-03T10:50:19.383937","created_date":"2023-12-11T00:00:00"}