{"id":"https://openalex.org/W7163227476","doi":"https://doi.org/10.48550/arxiv.2606.00424","title":"Weak Critics Make Strong Learners: On-Policy Critique Distillation for Scalable Oversight","display_name":"Weak Critics Make Strong Learners: On-Policy Critique Distillation for Scalable Oversight","publication_year":2026,"publication_date":"2026-05-29","ids":{"openalex":"https://openalex.org/W7163227476","doi":"https://doi.org/10.48550/arxiv.2606.00424"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2606.00424","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.00424","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":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.2606.00424","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5137670405","display_name":"Can Jin","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jin, Can","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5137623106","display_name":"Jiakang Li","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Li, Jiakang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5137653306","display_name":"Rui Wu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wu, Rui","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5069358248","display_name":"Eddy Zhang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhang, Eddy","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5137674592","display_name":"Dimitris N. Metaxas","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Metaxas, Dimitris N.","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/T10028","display_name":"Topic Modeling","score":0.37290000915527344,"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":0.37290000915527344,"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":0.14180000126361847,"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/T11714","display_name":"Multimodal Machine Learning Applications","score":0.08150000125169754,"subfield":{"id":"https://openalex.org/subfields/1707","display_name":"Computer Vision and Pattern Recognition"},"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/scalability","display_name":"Scalability","score":0.7537000179290771},{"id":"https://openalex.org/keywords/generalization","display_name":"Generalization","score":0.6814000010490417},{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.6488000154495239},{"id":"https://openalex.org/keywords/limiting","display_name":"Limiting","score":0.539900004863739},{"id":"https://openalex.org/keywords/key","display_name":"Key (lock)","score":0.5058000087738037},{"id":"https://openalex.org/keywords/task","display_name":"Task (project management)","score":0.48019999265670776},{"id":"https://openalex.org/keywords/path","display_name":"Path (computing)","score":0.46799999475479126},{"id":"https://openalex.org/keywords/quality","display_name":"Quality (philosophy)","score":0.4602000117301941}],"concepts":[{"id":"https://openalex.org/C48044578","wikidata":"https://www.wikidata.org/wiki/Q727490","display_name":"Scalability","level":2,"score":0.7537000179290771},{"id":"https://openalex.org/C177148314","wikidata":"https://www.wikidata.org/wiki/Q170084","display_name":"Generalization","level":2,"score":0.6814000010490417},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6552000045776367},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.6488000154495239},{"id":"https://openalex.org/C188198153","wikidata":"https://www.wikidata.org/wiki/Q1613840","display_name":"Limiting","level":2,"score":0.539900004863739},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.5058000087738037},{"id":"https://openalex.org/C2780451532","wikidata":"https://www.wikidata.org/wiki/Q759676","display_name":"Task (project management)","level":2,"score":0.48019999265670776},{"id":"https://openalex.org/C2777735758","wikidata":"https://www.wikidata.org/wiki/Q817765","display_name":"Path (computing)","level":2,"score":0.46799999475479126},{"id":"https://openalex.org/C2779530757","wikidata":"https://www.wikidata.org/wiki/Q1207505","display_name":"Quality (philosophy)","level":2,"score":0.4602000117301941},{"id":"https://openalex.org/C204030448","wikidata":"https://www.wikidata.org/wiki/Q101017","display_name":"Distillation","level":2,"score":0.36239999532699585},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.3621000051498413},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.3384999930858612},{"id":"https://openalex.org/C165064840","wikidata":"https://www.wikidata.org/wiki/Q1321061","display_name":"Matching (statistics)","level":2,"score":0.33550000190734863},{"id":"https://openalex.org/C98184364","wikidata":"https://www.wikidata.org/wiki/Q1780131","display_name":"Argument (complex analysis)","level":2,"score":0.30630001425743103},{"id":"https://openalex.org/C2778755073","wikidata":"https://www.wikidata.org/wiki/Q10858537","display_name":"Scale (ratio)","level":2,"score":0.2800999879837036},{"id":"https://openalex.org/C39920170","wikidata":"https://www.wikidata.org/wiki/Q693083","display_name":"Soundness","level":2,"score":0.2750999927520752},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.2709999978542328},{"id":"https://openalex.org/C77618280","wikidata":"https://www.wikidata.org/wiki/Q1155772","display_name":"Scheme (mathematics)","level":2,"score":0.2700999975204468},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.26109999418258667},{"id":"https://openalex.org/C144237770","wikidata":"https://www.wikidata.org/wiki/Q747534","display_name":"Mathematical economics","level":1,"score":0.25690001249313354},{"id":"https://openalex.org/C176248197","wikidata":"https://www.wikidata.org/wiki/Q458526","display_name":"Probably approximately correct learning","level":4,"score":0.2547999918460846}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2606.00424","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.00424","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:10.48550/arxiv.2606.00424","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.00424","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":false,"raw_source_name":null,"raw_type":"Preprint"},"sustainable_development_goals":[{"score":0.7379298806190491,"display_name":"Quality Education","id":"https://metadata.un.org/sdg/4"}],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"As":[0],"large":[1],"language":[2],"models":[3,99,148],"become":[4],"stronger,":[5],"weak":[6,35,39,62,93,160],"supervisors":[7],"may":[8],"fail":[9],"to":[10,66,109],"provide":[11,67],"reliable":[12],"labels,":[13],"preferences,":[14],"or":[15,49,56],"final":[16],"judgments":[17],"for":[18,156],"complex":[19],"outputs,":[20],"limiting":[21],"both":[22],"weak-to-strong":[23],"generalization":[24],"and":[25,103,124,139],"scalable":[26,157],"oversight.":[27],"We":[28,82,89,112],"study":[29],"a":[30,38,42,47,68],"more":[31],"tractable":[32],"form":[33],"of":[34,52],"supervision:":[36],"using":[37],"model":[40,76,131],"as":[41,46],"critic":[43,63],"rather":[44],"than":[45],"labeler":[48],"judge.":[50],"Instead":[51],"solving":[53],"the":[54,58,61,74,129],"task":[55],"selecting":[57],"correct":[59],"answer,":[60],"only":[64],"needs":[65],"non-misleading":[69],"revision":[70],"direction":[71],"that":[72,92,104,143],"helps":[73],"strong":[75,87,98,130,147],"better":[77],"use":[78],"its":[79],"own":[80],"knowledge.":[81],"call":[83],"this":[84,110],"setting":[85],"*weak-critic":[86],"oversight*.":[88],"first":[90],"show":[91,142],"critiques":[94,123],"can":[95],"improve":[96],"frozen":[97],"at":[100],"inference":[101],"time,":[102],"critique":[105,117],"quality":[106],"is":[107],"key":[108],"improvement.":[111],"then":[113],"propose":[114],"progressive":[115],"on-policy":[116],"distillation":[118],"(**OPCD**),":[119],"which":[120],"filters":[121],"high-quality":[122],"distills":[125],"critic-guided":[126],"behavior":[127],"into":[128],"through":[132],"adaptive":[133],"self-teacher":[134],"signals.":[135],"Experiments":[136],"on":[137],"reasoning":[138],"alignment":[140],"benchmarks":[141],"our":[144],"method":[145],"improves":[146],"over":[149],"training":[150],"epochs,":[151],"suggesting":[152],"an":[153],"effective":[154],"path":[155],"oversight":[158],"with":[159],"supervision.":[161]},"counts_by_year":[],"updated_date":"2026-07-28T07:46:37.118299","created_date":"2026-06-03T00:00:00"}