{"id":"https://openalex.org/W4415332749","doi":"https://doi.org/10.18653/v1/2026.acl-long.67","title":"J4R: Learning to Judge with Equivalent Initial State Group Relative Policy Optimization","display_name":"J4R: Learning to Judge with Equivalent Initial State Group Relative Policy Optimization","publication_year":2026,"publication_date":"2026-01-01","ids":{"openalex":"https://openalex.org/W4415332749","doi":"https://doi.org/10.18653/v1/2026.acl-long.67"},"language":"en","primary_location":{"id":"doi:10.18653/v1/2026.acl-long.67","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.acl-long.67","pdf_url":"https://aclanthology.org/2026.acl-long.67.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 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"},"type":"conference-paper","indexed_in":["arxiv","crossref","datacite"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://aclanthology.org/2026.acl-long.67.pdf","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5002266393","display_name":"Austin Xu","orcid":"https://orcid.org/0009-0003-1694-1841"},"institutions":[{"id":"https://openalex.org/I4210155268","display_name":"Salesforce (United States)","ror":"https://ror.org/057315g56","country_code":"US","type":"company","lineage":["https://openalex.org/I4210155268"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Austin Xu","raw_affiliation_strings":["Salesforce AI Research"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Salesforce AI Research","institution_ids":["https://openalex.org/I4210155268"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5053911077","display_name":"Yilun Zhou","orcid":"https://orcid.org/0000-0003-4759-4937"},"institutions":[{"id":"https://openalex.org/I4210155268","display_name":"Salesforce (United States)","ror":"https://ror.org/057315g56","country_code":"US","type":"company","lineage":["https://openalex.org/I4210155268"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Yilun Zhou","raw_affiliation_strings":["Salesforce AI Research"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Salesforce AI Research","institution_ids":["https://openalex.org/I4210155268"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5114373630","display_name":"Xuan-Phi Nguyen","orcid":null},"institutions":[{"id":"https://openalex.org/I4210155268","display_name":"Salesforce (United States)","ror":"https://ror.org/057315g56","country_code":"US","type":"company","lineage":["https://openalex.org/I4210155268"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Xuan-Phi Nguyen","raw_affiliation_strings":["Salesforce AI Research"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Salesforce AI Research","institution_ids":["https://openalex.org/I4210155268"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5032046813","display_name":"Caiming Xiong","orcid":"https://orcid.org/0000-0003-0349-8628"},"institutions":[{"id":"https://openalex.org/I4210155268","display_name":"Salesforce (United States)","ror":"https://ror.org/057315g56","country_code":"US","type":"company","lineage":["https://openalex.org/I4210155268"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Caiming Xiong","raw_affiliation_strings":["Salesforce AI Research"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Salesforce AI Research","institution_ids":["https://openalex.org/I4210155268"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5005443526","display_name":"Shafiq Joty","orcid":"https://orcid.org/0000-0002-9222-2641"},"institutions":[{"id":"https://openalex.org/I4210155268","display_name":"Salesforce (United States)","ror":"https://ror.org/057315g56","country_code":"US","type":"company","lineage":["https://openalex.org/I4210155268"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Shafiq Joty","raw_affiliation_strings":["Salesforce AI Research"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Salesforce AI Research","institution_ids":["https://openalex.org/I4210155268"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":["https://openalex.org/I4210155268"],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":true,"cited_by_count":0,"citation_normalized_percentile":{"value":0.00285987,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"1492","last_page":"1511"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T13643","display_name":"Artificial Intelligence in Law","score":0.8551999926567078,"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"}},"topics":[{"id":"https://openalex.org/T13643","display_name":"Artificial Intelligence in Law","score":0.8551999926567078,"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"}}],"keywords":[{"id":"https://openalex.org/keywords/pace","display_name":"Pace","score":0.7664999961853027},{"id":"https://openalex.org/keywords/benchmark","display_name":"Benchmark (surveying)","score":0.7433000206947327},{"id":"https://openalex.org/keywords/matching","display_name":"Matching (statistics)","score":0.6603000164031982},{"id":"https://openalex.org/keywords/reinforcement-learning","display_name":"Reinforcement learning","score":0.6446999907493591},{"id":"https://openalex.org/keywords/generative-grammar","display_name":"Generative grammar","score":0.5012999773025513},{"id":"https://openalex.org/keywords/state","display_name":"State (computer science)","score":0.4796999990940094},{"id":"https://openalex.org/keywords/class","display_name":"Class (philosophy)","score":0.4429999887943268}],"concepts":[{"id":"https://openalex.org/C2777526511","wikidata":"https://www.wikidata.org/wiki/Q691543","display_name":"Pace","level":2,"score":0.7664999961853027},{"id":"https://openalex.org/C185798385","wikidata":"https://www.wikidata.org/wiki/Q1161707","display_name":"Benchmark (surveying)","level":2,"score":0.7433000206947327},{"id":"https://openalex.org/C165064840","wikidata":"https://www.wikidata.org/wiki/Q1321061","display_name":"Matching (statistics)","level":2,"score":0.6603000164031982},{"id":"https://openalex.org/C97541855","wikidata":"https://www.wikidata.org/wiki/Q830687","display_name":"Reinforcement learning","level":2,"score":0.6446999907493591},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6175000071525574},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.555400013923645},{"id":"https://openalex.org/C39890363","wikidata":"https://www.wikidata.org/wiki/Q36108","display_name":"Generative grammar","level":2,"score":0.5012999773025513},{"id":"https://openalex.org/C48103436","wikidata":"https://www.wikidata.org/wiki/Q599031","display_name":"State (computer science)","level":2,"score":0.4796999990940094},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.46939998865127563},{"id":"https://openalex.org/C2777212361","wikidata":"https://www.wikidata.org/wiki/Q5127848","display_name":"Class (philosophy)","level":2,"score":0.4429999887943268},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.39259999990463257},{"id":"https://openalex.org/C2780586882","wikidata":"https://www.wikidata.org/wiki/Q7520643","display_name":"Simple (philosophy)","level":2,"score":0.38940000534057617},{"id":"https://openalex.org/C167966045","wikidata":"https://www.wikidata.org/wiki/Q5532625","display_name":"Generative model","level":3,"score":0.3709999918937683},{"id":"https://openalex.org/C177264268","wikidata":"https://www.wikidata.org/wiki/Q1514741","display_name":"Set (abstract data type)","level":2,"score":0.3345000147819519},{"id":"https://openalex.org/C2781311116","wikidata":"https://www.wikidata.org/wiki/Q83306","display_name":"Group (periodic table)","level":2,"score":0.31700000166893005},{"id":"https://openalex.org/C12725497","wikidata":"https://www.wikidata.org/wiki/Q810247","display_name":"Baseline (sea)","level":2,"score":0.2930999994277954},{"id":"https://openalex.org/C42475967","wikidata":"https://www.wikidata.org/wiki/Q194292","display_name":"Operations research","level":1,"score":0.28029999136924744},{"id":"https://openalex.org/C2781384469","wikidata":"https://www.wikidata.org/wiki/Q6786277","display_name":"Matching law","level":3,"score":0.2563000023365021},{"id":"https://openalex.org/C148220186","wikidata":"https://www.wikidata.org/wiki/Q7111912","display_name":"Outcome (game theory)","level":2,"score":0.2506999969482422}],"mesh":[],"locations_count":3,"locations":[{"id":"doi:10.18653/v1/2026.acl-long.67","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.acl-long.67","pdf_url":"https://aclanthology.org/2026.acl-long.67.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 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"},{"id":"pmh:oai:arXiv.org:2505.13346","is_oa":true,"landing_page_url":"http://arxiv.org/abs/2505.13346","pdf_url":"https://arxiv.org/pdf/2505.13346","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,"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.2505.13346","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2505.13346","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,"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":null,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"doi:10.18653/v1/2026.acl-long.67","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.acl-long.67","pdf_url":"https://aclanthology.org/2026.acl-long.67.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 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4415332749.pdf","grobid_xml":"https://content.openalex.org/works/W4415332749.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"To":[0],"keep":[1],"pace":[2],"with":[3,31,78,146],"the":[4,89,152,164],"increasing":[5],"velocity":[6],"of":[7,42,166,187],"large":[8],"language":[9],"models":[10,38],"(LLM)":[11],"development,":[12],"model":[13,36,62],"output":[14],"evaluation":[15,22,117,186],"has":[16],"transitioned":[17],"away":[18],"from":[19],"time-consuming":[20],"human":[21],"to":[23,102,106,109],"automatic":[24],"evaluation,":[25],"where":[26,61],"LLMs":[27],"themselves":[28],"are":[29,39],"tasked":[30],"assessing":[32],"and":[33,67,151,159,173],"critiquing":[34],"other":[35],"outputs.LLM-as-judge":[37],"a":[40,122,142,200],"class":[41],"generative":[43],"evaluators":[44],"that":[45,112,124,148],"excel":[46],"in":[47,57,114,127,189],"evaluating":[48],"relatively":[49],"simple":[50],"domains,":[51],"like":[52],"chat":[53],"quality,":[54],"but":[55],"struggle":[56],"reasoning":[58,129,190,204],"intensive":[59],"domains":[60],"responses":[63],"contain":[64],"more":[65,115],"substantive":[66],"challenging":[68],"content.To":[69],"remedy":[70],"existing":[71],"judge":[72,105,144,156,196],"shortcomings,":[73],"we":[74],"explore":[75],"training":[76,197],"judges":[77,126,169,188],"reinforcement":[79],"learning":[80],"(RL).We":[81],"make":[82],"three":[83],"key":[84],"contributions:":[85],"(1)":[86],"We":[87,119,136],"propose":[88],"Equivalent":[90],"Initial":[91],"State":[92],"Group":[93],"Relative":[94],"Policy":[95],"Optimization":[96],"(EIS-GRPO)":[97],"algorithm,":[98],"which":[99],"allows":[100],"us":[101],"train":[103,137],"our":[104],"be":[107],"robust":[108,195],"positional":[110],"biases":[111],"arise":[113],"complex":[116],"settings.(2)":[118],"introduce":[120],"ReasoningJudgeBench,":[121],"benchmark":[123],"evaluates":[125],"diverse":[128],"settings":[130,191],"not":[131],"covered":[132],"by":[133,157],"prior":[134],"work.(3)":[135],"Judge":[138],"for":[139,194],"Reasoning":[140],"(J4R),":[141],"7B":[143],"trained":[145],"EIS-GRPO":[147,193],"outperforms":[149],"GPT-4o":[150],"next":[153],"best":[154],"small":[155,201],"6.7%":[158],"9%,":[160],"matching":[161],"or":[162],"exceeding":[163],"performance":[165],"larger":[167],"GRPOtrained":[168],"on":[170],"both":[171],"JudgeBench":[172],"Rea-soningJudgeBench.":[174],"Reward":[175],"+":[176],"Advantage":[177],"Training":[178],"w/":[179],"transformation":[180],"invariant":[181],"RL":[182],"algorithm":[183],"Benchmark:":[184],"Enhanced":[185],"Methodology:":[192],"Model:":[198],"J4R-7B,":[199],"yet":[202],"effective":[203],"evaluator":[205]},"counts_by_year":[],"updated_date":"2026-08-05T07:39:15.569665","created_date":"2025-10-19T00:00:00"}