{"id":"https://openalex.org/W4385569890","doi":"https://doi.org/10.18653/v1/2023.acl-long.262","title":"Reward Gaming in Conditional Text Generation","display_name":"Reward Gaming in Conditional Text Generation","publication_year":2023,"publication_date":"2023-01-01","ids":{"openalex":"https://openalex.org/W4385569890","doi":"https://doi.org/10.18653/v1/2023.acl-long.262"},"language":"en","primary_location":{"id":"doi:10.18653/v1/2023.acl-long.262","doi":"https://doi.org/10.18653/v1/2023.acl-long.262","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2023.acl-long.262","pdf_url":"https://aclanthology.org/2023.acl-long.262.pdf","source":{"id":"https://openalex.org/S4393915740","display_name":"Proceedings of the conference - Association for Computational Linguistics. Meeting","issn_l":"0736-587X","issn":["0736-587X"],"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":["norway-1"],"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 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"},"volume":null,"type":"conference-paper","is_traditional":true,"indexed_in":["arxiv","crossref","datacite"],"open_access":{"is_oa":true,"oa_status":"hybrid","oa_url":"https://aclanthology.org/2023.acl-long.262.pdf","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5103936332","display_name":"Richard Yuanzhe Pang","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I57206974","display_name":"New York University","ror":"https://ror.org/0190ak572","country_code":"US","type":"education","lineage":["https://openalex.org/I57206974"]}],"countries":["US"],"is_corresponding":true,"raw_author_name":"Richard Yuanzhe Pang","raw_affiliation_strings":["New York University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"New York University","institution_ids":["https://openalex.org/I57206974"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5052649262","display_name":"Vishakh Padmakumar","orcid":"https://orcid.org/0000-0002-3396-3589","observed_orcids":["https://orcid.org/0000-0002-3396-3589"]},"institutions":[{"id":"https://openalex.org/I57206974","display_name":"New York University","ror":"https://ror.org/0190ak572","country_code":"US","type":"education","lineage":["https://openalex.org/I57206974"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Vishakh Padmakumar","raw_affiliation_strings":["New York University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"New York University","institution_ids":["https://openalex.org/I57206974"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5069698101","display_name":"Thibault Sellam","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I4210090411","display_name":"Google DeepMind (United Kingdom)","ror":"https://ror.org/00971b260","country_code":"GB","type":"company","lineage":["https://openalex.org/I4210090411","https://openalex.org/I4210128969"]}],"countries":["GB"],"is_corresponding":false,"raw_author_name":"Thibault Sellam","raw_affiliation_strings":["Google DeepMind"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Google DeepMind","institution_ids":["https://openalex.org/I4210090411"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5109937799","display_name":"Ankur P. Parikh","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I4210090411","display_name":"Google DeepMind (United Kingdom)","ror":"https://ror.org/00971b260","country_code":"GB","type":"company","lineage":["https://openalex.org/I4210090411","https://openalex.org/I4210128969"]}],"countries":["GB"],"is_corresponding":false,"raw_author_name":"Ankur Parikh","raw_affiliation_strings":["Google DeepMind"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Google DeepMind","institution_ids":["https://openalex.org/I4210090411"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5100775952","display_name":"He He","orcid":"https://orcid.org/0000-0002-9118-2449","observed_orcids":["https://orcid.org/0000-0002-9118-2449","https://orcid.org/0009-0005-7805-3780"]},"institutions":[{"id":"https://openalex.org/I57206974","display_name":"New York University","ror":"https://ror.org/0190ak572","country_code":"US","type":"education","lineage":["https://openalex.org/I57206974"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"He He","raw_affiliation_strings":["New York University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"New York University","institution_ids":["https://openalex.org/I57206974"]}]}],"institutions":[],"countries_distinct_count":2,"institutions_distinct_count":2,"corresponding_author_ids":["https://openalex.org/A5103936332"],"corresponding_institution_ids":["https://openalex.org/I57206974"],"apc_list":null,"apc_paid":null,"fwci":2.4304,"has_fulltext":true,"cited_by_count":22,"citation_normalized_percentile":{"value":0.92342893,"is_in_top_1_percent":false,"is_in_top_10_percent":true},"cited_by_percentile_year":{"min":93,"max":99},"biblio":{"volume":null,"issue":null,"first_page":"4746","last_page":"4763"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10181","display_name":"Natural Language Processing Techniques","score":0.47269999980926514,"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":0.47269999980926514,"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":0.33959999680519104,"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":0.03970000147819519,"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/conditional-text-generation","display_name":"conditional text generation","score":0.9470000267028809},{"id":"https://openalex.org/keywords/reinforcement-learning","display_name":"reinforcement learning","score":0.6190000176429749},{"id":"https://openalex.org/keywords/reward-function","display_name":"reward function","score":0.7559999823570251},{"id":"https://openalex.org/keywords/spurious-correlations","display_name":"spurious correlations","score":0.8999999761581421},{"id":"https://openalex.org/keywords/covariate-shift","display_name":"covariate shift","score":0.9100000262260437},{"id":"https://openalex.org/keywords/natural-language-generation","display_name":"natural language generation","score":0.777999997138977}],"concepts":[{"id":"https://openalex.org/C97256817","wikidata":"https://www.wikidata.org/wiki/Q1462316","display_name":"Spurious relationship","level":2,"score":0.8941196203231812},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7739481925964355},{"id":"https://openalex.org/C97541855","wikidata":"https://www.wikidata.org/wiki/Q830687","display_name":"Reinforcement learning","level":2,"score":0.7353057861328125},{"id":"https://openalex.org/C119043178","wikidata":"https://www.wikidata.org/wiki/Q320723","display_name":"Covariate","level":2,"score":0.6003100872039795},{"id":"https://openalex.org/C2985684807","wikidata":"https://www.wikidata.org/wiki/Q1513879","display_name":"Text generation","level":2,"score":0.5991665720939636},{"id":"https://openalex.org/C192209626","wikidata":"https://www.wikidata.org/wiki/Q190909","display_name":"Focus (optics)","level":2,"score":0.55074542760849},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.541883111000061},{"id":"https://openalex.org/C14036430","wikidata":"https://www.wikidata.org/wiki/Q3736076","display_name":"Function (biology)","level":2,"score":0.4684220552444458},{"id":"https://openalex.org/C2776187449","wikidata":"https://www.wikidata.org/wiki/Q1513879","display_name":"Natural language generation","level":3,"score":0.4603028893470764},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.46022942662239075},{"id":"https://openalex.org/C195324797","wikidata":"https://www.wikidata.org/wiki/Q33742","display_name":"Natural language","level":2,"score":0.2842881381511688},{"id":"https://openalex.org/C120665830","wikidata":"https://www.wikidata.org/wiki/Q14620","display_name":"Optics","level":1,"score":0.0},{"id":"https://openalex.org/C86803240","wikidata":"https://www.wikidata.org/wiki/Q420","display_name":"Biology","level":0,"score":0.0},{"id":"https://openalex.org/C121332964","wikidata":"https://www.wikidata.org/wiki/Q413","display_name":"Physics","level":0,"score":0.0},{"id":"https://openalex.org/C78458016","wikidata":"https://www.wikidata.org/wiki/Q840400","display_name":"Evolutionary biology","level":1,"score":0.0}],"mesh":[],"locations_count":3,"locations":[{"id":"doi:10.18653/v1/2023.acl-long.262","doi":"https://doi.org/10.18653/v1/2023.acl-long.262","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2023.acl-long.262","pdf_url":"https://aclanthology.org/2023.acl-long.262.pdf","source":{"id":"https://openalex.org/S4393915740","display_name":"Proceedings of the conference - Association for Computational Linguistics. Meeting","issn_l":"0736-587X","issn":["0736-587X"],"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":["norway-1"],"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 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"},{"id":"pmh:oai:arXiv.org:2211.08714","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2211.08714","pdf_url":"https://arxiv.org/pdf/2211.08714","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"},{"id":"doi:10.48550/arxiv.2211.08714","doi":"https://doi.org/10.48550/arxiv.2211.08714","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2211.08714","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":"Preprint"}],"best_oa_location":{"id":"doi:10.18653/v1/2023.acl-long.262","doi":"https://doi.org/10.18653/v1/2023.acl-long.262","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2023.acl-long.262","pdf_url":"https://aclanthology.org/2023.acl-long.262.pdf","source":{"id":"https://openalex.org/S4393915740","display_name":"Proceedings of the conference - Association for Computational Linguistics. Meeting","issn_l":"0736-587X","issn":["0736-587X"],"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":["norway-1"],"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 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"},"sustainable_development_goals":[],"study_designs":[],"awards":[],"funders":[{"id":"https://openalex.org/F4320315121","display_name":"Samsung Advanced Institute of Technology","ror":"https://ror.org/04axnyp10"},{"id":"https://openalex.org/F4320332195","display_name":"Samsung","ror":"https://ror.org/04w3jy968"}],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4385569890.pdf","grobid_xml":"https://content.openalex.org/works/W4385569890.grobid-xml"},"referenced_works_count":77,"referenced_works":["https://openalex.org/W648786980","https://openalex.org/W1522301498","https://openalex.org/W1654441844","https://openalex.org/W2119717200","https://openalex.org/W2155027007","https://openalex.org/W2174424190","https://openalex.org/W2176263492","https://openalex.org/W2462906003","https://openalex.org/W2508316494","https://openalex.org/W2508920525","https://openalex.org/W2525778437","https://openalex.org/W2625113742","https://openalex.org/W2736601468","https://openalex.org/W2795776994","https://openalex.org/W2797269287","https://openalex.org/W2889659926","https://openalex.org/W2903188467","https://openalex.org/W2915756181","https://openalex.org/W2916548775","https://openalex.org/W2933138175","https://openalex.org/W2949678053","https://openalex.org/W2953981431","https://openalex.org/W2963248348","https://openalex.org/W2964121744","https://openalex.org/W2965373594","https://openalex.org/W2968297680","https://openalex.org/W2968771440","https://openalex.org/W2970279348","https://openalex.org/W2970692082","https://openalex.org/W2970791445","https://openalex.org/W2973525135","https://openalex.org/W2975835314","https://openalex.org/W2979214993","https://openalex.org/W2982399380","https://openalex.org/W2997502221","https://openalex.org/W3013891696","https://openalex.org/W3015475346","https://openalex.org/W3033324992","https://openalex.org/W3082115681","https://openalex.org/W3085932930","https://openalex.org/W3100715140","https://openalex.org/W3104608349","https://openalex.org/W3119775721","https://openalex.org/W3119878165","https://openalex.org/W3120929527","https://openalex.org/W3120979723","https://openalex.org/W3124120397","https://openalex.org/W3148330722","https://openalex.org/W3159892921","https://openalex.org/W3165946561","https://openalex.org/W3172079109","https://openalex.org/W3174785103","https://openalex.org/W3175563972","https://openalex.org/W3196392173","https://openalex.org/W3200980294","https://openalex.org/W3201464480","https://openalex.org/W3204130541","https://openalex.org/W3213632785","https://openalex.org/W4223908421","https://openalex.org/W4226151089","https://openalex.org/W4226251505","https://openalex.org/W4285112711","https://openalex.org/W4302305863","https://openalex.org/W4307001524","https://openalex.org/W4308242490","https://openalex.org/W4311991106","https://openalex.org/W4312050653","https://openalex.org/W4320013936","https://openalex.org/W4361866126","https://openalex.org/W4385570669","https://openalex.org/W4385571099","https://openalex.org/W4385571158","https://openalex.org/W4385573519","https://openalex.org/W4389009525","https://openalex.org/W4404783892","https://openalex.org/W4404783940","https://openalex.org/W7133230429"],"related_works":["https://openalex.org/W7133204914","https://openalex.org/W3115418361","https://openalex.org/W4389523656","https://openalex.org/W7166808858","https://openalex.org/W2953141401","https://openalex.org/W4389519204","https://openalex.org/W4399694253","https://openalex.org/W4407308896","https://openalex.org/W3022309035","https://openalex.org/W4412888857"],"abstract_inverted_index":{"To":[0],"align":[1],"conditional":[2,125],"text":[3,89,126],"generation":[4,90,120,127],"model":[5,19],"outputs":[6],"with":[7,24],"desired":[8],"behaviors,":[9],"there":[10,92],"has":[11,93],"been":[12,94],"an":[13],"increasing":[14],"focus":[15],"on":[16,66],"training":[17,86],"the":[18,67,70,75,78,88,100,117],"using":[20,123],"reinforcement":[21],"learning":[22],"(RL)":[23],"reward":[25,76,97,114],"functions":[26],"learned":[27,61],"from":[28],"human":[29],"annotations.Under":[30],"this":[31,106],"framework,":[32],"we":[33,109],"identify":[34],"three":[35],"common":[36],"cases":[37],"where":[38],"high":[39,64],"rewards":[40],"are":[41],"incorrectly":[42],"assigned":[43],"to":[44,73,112],"undesirable":[45,79],"patterns:":[46],"noise-induced":[47],"spurious":[48,52],"correlation,":[49,53],"naturally":[50],"occurring":[51],"and":[54,129,133],"covariate":[55],"shift.We":[56],"show":[57],"that":[58],"even":[59],"though":[60],"metrics":[62],"achieve":[63],"performance":[65],"distribution":[68],"of":[69,87],"data":[71],"used":[72],"train":[74],"function,":[77],"patterns":[80],"may":[81],"be":[82],"amplified":[83],"during":[84],"RL":[85,101],"model.While":[91],"discussion":[95,107],"about":[96],"gaming":[98,115],"in":[99,105,116],"or":[102],"safety":[103],"community,":[104],"piece,":[108],"would":[110],"like":[111],"highlight":[113],"natural":[118],"language":[119],"(NLG)":[121],"community":[122],"concrete":[124],"examples":[128],"discuss":[130],"potential":[131],"fixes":[132],"areas":[134],"for":[135],"future":[136],"work.":[137]},"counts_by_year":[{"year":2026,"cited_by_count":4},{"year":2025,"cited_by_count":8},{"year":2024,"cited_by_count":8},{"year":2023,"cited_by_count":2}],"updated_date":"2026-11-10T11:39:43.021295","created_date":"2023-08-05T00:00:00"}