{"id":"https://openalex.org/W7225834039","doi":"https://doi.org/10.48550/arxiv.2611.13680","title":"Bootstrapped On-Policy Distillation","display_name":"Bootstrapped On-Policy Distillation","publication_year":2026,"publication_date":"2026-11-10","ids":{"openalex":"https://openalex.org/W7225834039","doi":"https://doi.org/10.48550/arxiv.2611.13680"},"language":"en","primary_location":{"id":"doi:10.48550/arxiv.2611.13680","doi":"https://doi.org/10.48550/arxiv.2611.13680","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2611.13680","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":"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"},"volume":null,"type":"preprint","is_traditional":true,"indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.48550/arxiv.2611.13680","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5161455114","display_name":"Yue Wu","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wu, Yue","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5129723060","display_name":"Xueyu Zhou","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhou, Xueyu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5129645870","display_name":"Yangrong Hu","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Hu, Yangrong","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5161434679","display_name":"Jian Huang","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Huang, Jian","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":0.7670999765396118,"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":0.7670999765396118,"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":0.08839999884366989,"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":0.08540000021457672,"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":[],"concepts":[{"id":"https://openalex.org/C153083717","wikidata":"https://www.wikidata.org/wiki/Q6535263","display_name":"Leverage (statistics)","level":2,"score":0.7246999740600586},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.682699978351593},{"id":"https://openalex.org/C204030448","wikidata":"https://www.wikidata.org/wiki/Q101017","display_name":"Distillation","level":2,"score":0.4388999938964844},{"id":"https://openalex.org/C2776760102","wikidata":"https://www.wikidata.org/wiki/Q5139990","display_name":"Code (set theory)","level":3,"score":0.4205999970436096},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.3650999963283539},{"id":"https://openalex.org/C77618280","wikidata":"https://www.wikidata.org/wiki/Q1155772","display_name":"Scheme (mathematics)","level":2,"score":0.35580000281333923},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.3151000142097473},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.3084999918937683},{"id":"https://openalex.org/C2777655017","wikidata":"https://www.wikidata.org/wiki/Q1501161","display_name":"Toolbox","level":2,"score":0.29910001158714294},{"id":"https://openalex.org/C179518139","wikidata":"https://www.wikidata.org/wiki/Q5140297","display_name":"Coding (social sciences)","level":2,"score":0.2948000133037567},{"id":"https://openalex.org/C21880701","wikidata":"https://www.wikidata.org/wiki/Q2144042","display_name":"Process engineering","level":1,"score":0.2847000062465668},{"id":"https://openalex.org/C174998907","wikidata":"https://www.wikidata.org/wiki/Q357662","display_name":"Work in process","level":2,"score":0.2759999930858612},{"id":"https://openalex.org/C76969082","wikidata":"https://www.wikidata.org/wiki/Q486902","display_name":"Mathematical model","level":2,"score":0.2614000141620636},{"id":"https://openalex.org/C89611455","wikidata":"https://www.wikidata.org/wiki/Q6804646","display_name":"Mechanism (biology)","level":2,"score":0.2563999891281128},{"id":"https://openalex.org/C112930515","wikidata":"https://www.wikidata.org/wiki/Q4389547","display_name":"Risk analysis (engineering)","level":1,"score":0.25619998574256897},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.2556999921798706},{"id":"https://openalex.org/C18762648","wikidata":"https://www.wikidata.org/wiki/Q42213","display_name":"Work (physics)","level":2,"score":0.25130000710487366},{"id":"https://openalex.org/C133462117","wikidata":"https://www.wikidata.org/wiki/Q4929239","display_name":"Data collection","level":2,"score":0.2502000033855438}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2611.13680","doi":"https://doi.org/10.48550/arxiv.2611.13680","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2611.13680","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":"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"}],"best_oa_location":{"id":"doi:10.48550/arxiv.2611.13680","doi":"https://doi.org/10.48550/arxiv.2611.13680","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2611.13680","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":"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":[],"study_designs":[],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"On-policy":[0],"distillation":[1],"(OPD)":[2],"improves":[3,201],"language":[4],"models":[5,197,211],"by":[6,24,229],"distilling":[7,241],"teacher":[8,23,90,118,125],"predictions":[9],"on":[10],"student-generated":[11],"trajectories,":[12],"while":[13],"on-policy":[14],"self-distillation":[15],"(OPSD)":[16],"removes":[17],"the":[18,30,33,66,73,78,83,88,95,103,111,116,124,192,209,213],"need":[19],"for":[20,101],"an":[21,177],"external":[22],"using":[25],"a":[26,69,153,162,188,236],"reference-conditioned":[27,174],"copy":[28,190],"of":[29,86,191],"model":[31,186,238,244],"as":[32,108],"teacher.":[34,81],"Despite":[35],"its":[36,232],"great":[37],"early":[38,84],"improvements,":[39],"we":[40],"find":[41],"that":[42,165,199,220,243],"OPSD":[43,168,215],"does":[44],"not":[45],"remain":[46],"consistently":[47],"effective":[48],"under":[49],"continued":[50],"training":[51,62,109],"and":[52,64,77,97,119,148,180,204,212,239],"can":[53,122,224],"eventually":[54],"suffer":[55],"substantial":[56],"performance":[57],"degradation.":[58],"We":[59,156],"study":[60],"this":[61,183],"behavior":[63],"link":[65],"degradation":[67],"to":[68,126,138,144,152,169],"growing":[70,112],"mismatch":[71,114],"between":[72,115],"evolving":[74],"reference-free":[75,120,184],"student":[76,96,121,132],"frozen":[79],"privileged":[80,89,117,141,221],"In":[82],"stage":[85],"OPSD,":[87],"remains":[91],"closely":[92],"aligned":[93],"with":[94],"provides":[98],"useful":[99,146],"guidance":[100],"improving":[102],"student's":[104],"generation":[105,206],"trajectories.":[106],"However,":[107],"progresses,":[110],"distributional":[113],"cause":[123],"provide":[127],"misleading":[128],"supervision,":[129],"ultimately":[130],"degrading":[131],"performance.":[133],"This":[134],"phenomenon":[135],"motivates":[136],"us":[137],"first":[139,166,230],"leverage":[140],"reference":[142,222,246],"supervision":[143,175,223],"acquire":[145],"knowledge":[147],"then":[149,181,240],"transfer":[150],"it":[151],"fresh":[154,189],"student.":[155],"introduce":[157],"Bootstrapped":[158],"On-Policy":[159],"Distillation":[160],"(BOPD),":[161],"two-stage":[163],"framework":[164],"uses":[167],"internalize":[170],"task-relevant":[171,233],"information":[172],"from":[173,242],"into":[176,187,235],"intermediate":[178,185],"model,":[179],"distills":[182],"initial":[193,210],"model.":[194],"Experiments":[195],"across":[196],"show":[198],"BOPD":[200],"mathematical":[202],"reasoning":[203],"code":[205],"beyond":[207],"both":[208],"constructed":[214],"teachers.":[216],"These":[217],"results":[218],"suggest":[219],"be":[225],"more":[226],"effectively":[227],"exploited":[228],"transferring":[231],"signal":[234],"task-adapted":[237],"without":[245],"conditioning.":[247]},"counts_by_year":[],"updated_date":"2026-11-12T12:28:57.059296","created_date":"2026-11-12T00:00:00"}