{"id":"https://openalex.org/W7160378838","doi":"https://doi.org/10.48550/arxiv.2605.03871","title":"EvoLM: Self-Evolving Language Models through Co-Evolved Discriminative Rubrics","display_name":"EvoLM: Self-Evolving Language Models through Co-Evolved Discriminative Rubrics","publication_year":2026,"publication_date":"2026-05-05","ids":{"openalex":"https://openalex.org/W7160378838","doi":"https://doi.org/10.48550/arxiv.2605.03871"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2605.03871","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.03871","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"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":false,"raw_source_name":null,"raw_type":"Preprint"},"type":"preprint","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.48550/arxiv.2605.03871","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5022934996","display_name":"Shuyue Stella Li","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Li, Shuyue Stella","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5135464069","display_name":"Rui Xin","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Xin, Rui","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5135485840","display_name":"Teng Xiao","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Xiao, Teng","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5101678590","display_name":"Yike Wang","orcid":"https://orcid.org/0009-0005-9765-8742"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wang, Yike","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5047742348","display_name":"Rulin Shao","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Shao, Rulin","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5135459091","display_name":"Zoey Hao","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Hao, Zoey","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5039472047","display_name":"Melanie Sclar","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Sclar, Melanie","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5135504437","display_name":"Sewoong Oh","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Oh, Sewoong","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5135471576","display_name":"Faeze Brahman","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Brahman, Faeze","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5135484356","display_name":"Pang Wei Koh","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Koh, Pang Wei","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5135440894","display_name":"Yulia Tsvetkov","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Tsvetkov, Yulia","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":null,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":null,"last_page":null},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11636","display_name":"Artificial Intelligence in Healthcare and Education","score":0.2632000148296356,"subfield":{"id":"https://openalex.org/subfields/2718","display_name":"Health Informatics"},"field":{"id":"https://openalex.org/fields/27","display_name":"Medicine"},"domain":{"id":"https://openalex.org/domains/4","display_name":"Health Sciences"}},"topics":[{"id":"https://openalex.org/T11636","display_name":"Artificial Intelligence in Healthcare and Education","score":0.2632000148296356,"subfield":{"id":"https://openalex.org/subfields/2718","display_name":"Health Informatics"},"field":{"id":"https://openalex.org/fields/27","display_name":"Medicine"},"domain":{"id":"https://openalex.org/domains/4","display_name":"Health Sciences"}},{"id":"https://openalex.org/T10028","display_name":"Topic Modeling","score":0.2304999977350235,"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":0.09399999678134918,"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/rubric","display_name":"Rubric","score":0.9110000133514404},{"id":"https://openalex.org/keywords/discriminative-model","display_name":"Discriminative model","score":0.8844000101089478},{"id":"https://openalex.org/keywords/language-model","display_name":"Language model","score":0.529699981212616},{"id":"https://openalex.org/keywords/encode","display_name":"ENCODE","score":0.43810001015663147},{"id":"https://openalex.org/keywords/vocabulary","display_name":"Vocabulary","score":0.4366999864578247},{"id":"https://openalex.org/keywords/generator","display_name":"Generator (circuit theory)","score":0.4230000078678131},{"id":"https://openalex.org/keywords/grammar","display_name":"Grammar","score":0.37630000710487366}],"concepts":[{"id":"https://openalex.org/C111640148","wikidata":"https://www.wikidata.org/wiki/Q847349","display_name":"Rubric","level":2,"score":0.9110000133514404},{"id":"https://openalex.org/C97931131","wikidata":"https://www.wikidata.org/wiki/Q5282087","display_name":"Discriminative model","level":2,"score":0.8844000101089478},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.684499979019165},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.536899983882904},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.529699981212616},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.4406000077724457},{"id":"https://openalex.org/C66746571","wikidata":"https://www.wikidata.org/wiki/Q1134833","display_name":"ENCODE","level":3,"score":0.43810001015663147},{"id":"https://openalex.org/C2777601683","wikidata":"https://www.wikidata.org/wiki/Q6499736","display_name":"Vocabulary","level":2,"score":0.4366999864578247},{"id":"https://openalex.org/C2780992000","wikidata":"https://www.wikidata.org/wiki/Q17016113","display_name":"Generator (circuit theory)","level":3,"score":0.4230000078678131},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.41819998621940613},{"id":"https://openalex.org/C26022165","wikidata":"https://www.wikidata.org/wiki/Q8091","display_name":"Grammar","level":2,"score":0.37630000710487366},{"id":"https://openalex.org/C190839683","wikidata":"https://www.wikidata.org/wiki/Q2448197","display_name":"Train","level":2,"score":0.3571000099182129},{"id":"https://openalex.org/C2775945657","wikidata":"https://www.wikidata.org/wiki/Q381442","display_name":"Structuring","level":2,"score":0.2849000096321106},{"id":"https://openalex.org/C55439883","wikidata":"https://www.wikidata.org/wiki/Q360812","display_name":"Correctness","level":2,"score":0.2797999978065491},{"id":"https://openalex.org/C2776187449","wikidata":"https://www.wikidata.org/wiki/Q1513879","display_name":"Natural language generation","level":3,"score":0.2718999981880188},{"id":"https://openalex.org/C2781249084","wikidata":"https://www.wikidata.org/wiki/Q908656","display_name":"Preference","level":2,"score":0.26109999418258667},{"id":"https://openalex.org/C2780428219","wikidata":"https://www.wikidata.org/wiki/Q16952335","display_name":"Cover (algebra)","level":2,"score":0.26030001044273376},{"id":"https://openalex.org/C22367795","wikidata":"https://www.wikidata.org/wiki/Q7625208","display_name":"Structured prediction","level":2,"score":0.25690001249313354}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2605.03871","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.03871","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"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:10.48550/arxiv.2605.03871","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.03871","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"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":false,"raw_source_name":null,"raw_type":"Preprint"},"sustainable_development_goals":[{"display_name":"Reduced inequalities","score":0.7550445199012756,"id":"https://metadata.un.org/sdg/10"}],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Language":[0],"models":[1],"encode":[2],"substantial":[3],"evaluative":[4,58,218],"knowledge":[5],"from":[6,54,130,149],"pretraining,":[7],"yet":[8,70],"current":[9,75],"post-training":[10,81],"methods":[11],"rely":[12],"on":[13,178,188],"external":[14,165,227],"supervision":[15],"(human":[16],"annotations,":[17],"proprietary":[18,40],"models,":[19],"or":[20,164],"scalar":[21],"reward":[22,26,62,206],"models)":[23],"to":[24,127,172],"produce":[25],"signals.":[27],"Each":[28],"imposes":[29],"a":[30,55,61,80,102,109,122,135,169,216],"ceiling.":[31],"Human":[32],"judgment":[33],"cannot":[34],"supervise":[35],"capabilities":[36,100],"beyond":[37],"its":[38],"own,":[39],"APIs":[41],"create":[42],"dependencies,":[43],"and":[44,91,133,201],"verifiable":[45],"rewards":[46],"cover":[47],"only":[48],"domains":[49],"with":[50,66,157,195,202],"ground-truth":[51],"answers.":[52],"Self-improvement":[53],"model's":[56,217],"own":[57,152],"capacity":[59,86,219],"is":[60],"source":[63],"that":[64,83,175,214],"scales":[65],"the":[67,150,189,203],"model":[68,105,171,207],"itself,":[69],"remains":[71],"largely":[72],"untapped":[73],"by":[74,180,199,209],"methods.":[76],"We":[77],"introduce":[78],"EVOLM,":[79],"method":[82],"structures":[84],"this":[85],"into":[87,220],"explicit":[88],"discriminative":[89,118,222],"rubrics":[90,174,198,223],"uses":[92],"them":[93],"as":[94,142],"training":[95],"signal.":[96],"EVOLM":[97,167,212],"trains":[98,168],"two":[99],"within":[101],"single":[103],"language":[104],"in":[106],"alternation:":[107],"(1)":[108],"rubric":[110],"generator":[111],"producing":[112],"instance-specific":[113],"evaluation":[114],"criteria":[115],"optimized":[116],"for":[117],"utility,":[119],"which":[120],"maximizes":[121],"small":[123],"frozen":[124],"judge's":[125],"ability":[126],"distinguish":[128],"preferred":[129],"dispreferred":[131],"responses;":[132],"(2)":[134],"policy":[136,184],"trained":[137,194],"using":[138],"those":[139],"rubric-conditioned":[140],"scores":[141],"reward.":[143],"All":[144],"preference":[145],"signals":[146],"are":[147],"constructed":[148],"policy's":[151],"outputs":[153],"via":[154],"temporal":[155],"contrast":[156],"earlier":[158],"checkpoints,":[159],"requiring":[160],"no":[161],"human":[162],"annotation":[163],"supervision.":[166,228],"Qwen3-8B":[170],"generate":[173],"outperform":[176],"GPT-4.1":[177,196],"RewardBench-2":[179],"25.7%.":[181],"The":[182],"co-trained":[183],"achieves":[185],"69.3%":[186],"average":[187],"OLMo3-Adapt":[190],"suite,":[191],"outperforming":[192],"policies":[193],"prompted":[197],"3.9%":[200],"state-of-the-art":[204],"8B":[205],"SkyWork-RM":[208],"16%.":[210],"Overall,":[211],"demonstrates":[213],"structuring":[215],"co-evolving":[221],"enables":[224],"self-improvement":[225],"without":[226]},"counts_by_year":[],"updated_date":"2026-07-01T06:00:48.157686","created_date":"2026-05-07T00:00:00"}