{"id":"https://openalex.org/W7160987778","doi":"https://doi.org/10.48550/arxiv.2605.11613","title":"From Generic Correlation to Input-Specific Credit in On-Policy Self Distillation","display_name":"From Generic Correlation to Input-Specific Credit in On-Policy Self Distillation","publication_year":2026,"publication_date":"2026-05-12","ids":{"openalex":"https://openalex.org/W7160987778","doi":"https://doi.org/10.48550/arxiv.2605.11613"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2605.11613","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.11613","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":null,"license_id":null,"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.2605.11613","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5081220735","display_name":"Guobin Shen","orcid":"https://orcid.org/0000-0002-4069-2107"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Shen, Guobin","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5136057104","display_name":"Lei Huang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Huang, Lei","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5136027578","display_name":"Xiang Cheng","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Cheng, Xiang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5136079920","display_name":"Chenxiao Zhao","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhao, Chenxiao","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5136014732","display_name":"Jindong Li","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Li, Jindong","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5136015040","display_name":"Dongcheng Zhao","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhao, Dongcheng","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5136050058","display_name":"Xing Yu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yu, Xing","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":0.3190999925136566,"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":0.3190999925136566,"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/T12031","display_name":"Speech and dialogue systems","score":0.10570000112056732,"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":0.08380000293254852,"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/pointwise","display_name":"Pointwise","score":0.6277999877929688},{"id":"https://openalex.org/keywords/component","display_name":"Component (thermodynamics)","score":0.5684000253677368},{"id":"https://openalex.org/keywords/sequence","display_name":"Sequence (biology)","score":0.501800000667572},{"id":"https://openalex.org/keywords/interpretation","display_name":"Interpretation (philosophy)","score":0.4952999949455261},{"id":"https://openalex.org/keywords/correlation","display_name":"Correlation","score":0.4828000068664551},{"id":"https://openalex.org/keywords/aggregate","display_name":"Aggregate (composite)","score":0.43939998745918274},{"id":"https://openalex.org/keywords/trajectory","display_name":"Trajectory","score":0.42260000109672546},{"id":"https://openalex.org/keywords/bayesian-inference","display_name":"Bayesian inference","score":0.414900004863739}],"concepts":[{"id":"https://openalex.org/C2777984123","wikidata":"https://www.wikidata.org/wiki/Q9248237","display_name":"Pointwise","level":2,"score":0.6277999877929688},{"id":"https://openalex.org/C168167062","wikidata":"https://www.wikidata.org/wiki/Q1117970","display_name":"Component (thermodynamics)","level":2,"score":0.5684000253677368},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5383999943733215},{"id":"https://openalex.org/C2778112365","wikidata":"https://www.wikidata.org/wiki/Q3511065","display_name":"Sequence (biology)","level":2,"score":0.501800000667572},{"id":"https://openalex.org/C527412718","wikidata":"https://www.wikidata.org/wiki/Q855395","display_name":"Interpretation (philosophy)","level":2,"score":0.4952999949455261},{"id":"https://openalex.org/C117220453","wikidata":"https://www.wikidata.org/wiki/Q5172842","display_name":"Correlation","level":2,"score":0.4828000068664551},{"id":"https://openalex.org/C149782125","wikidata":"https://www.wikidata.org/wiki/Q160039","display_name":"Econometrics","level":1,"score":0.4580000042915344},{"id":"https://openalex.org/C4679612","wikidata":"https://www.wikidata.org/wiki/Q866298","display_name":"Aggregate (composite)","level":2,"score":0.43939998745918274},{"id":"https://openalex.org/C13662910","wikidata":"https://www.wikidata.org/wiki/Q193139","display_name":"Trajectory","level":2,"score":0.42260000109672546},{"id":"https://openalex.org/C160234255","wikidata":"https://www.wikidata.org/wiki/Q812535","display_name":"Bayesian inference","level":3,"score":0.414900004863739},{"id":"https://openalex.org/C176217482","wikidata":"https://www.wikidata.org/wiki/Q860554","display_name":"Metric (unit)","level":2,"score":0.4146000146865845},{"id":"https://openalex.org/C107673813","wikidata":"https://www.wikidata.org/wiki/Q812534","display_name":"Bayesian probability","level":2,"score":0.40209999680519104},{"id":"https://openalex.org/C120936955","wikidata":"https://www.wikidata.org/wiki/Q2155640","display_name":"Empirical research","level":2,"score":0.38920000195503235},{"id":"https://openalex.org/C7797323","wikidata":"https://www.wikidata.org/wiki/Q3798612","display_name":"Pointwise mutual information","level":3,"score":0.3824999928474426},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.32910001277923584},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.2957000136375427},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.2847000062465668},{"id":"https://openalex.org/C152139883","wikidata":"https://www.wikidata.org/wiki/Q252973","display_name":"Mutual information","level":2,"score":0.27570000290870667},{"id":"https://openalex.org/C162324750","wikidata":"https://www.wikidata.org/wiki/Q8134","display_name":"Economics","level":0,"score":0.2685000002384186},{"id":"https://openalex.org/C2775924081","wikidata":"https://www.wikidata.org/wiki/Q55608371","display_name":"Control (management)","level":2,"score":0.2646999955177307},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.2606000006198883},{"id":"https://openalex.org/C2780009758","wikidata":"https://www.wikidata.org/wiki/Q6804172","display_name":"Measure (data warehouse)","level":2,"score":0.2526000142097473}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2605.11613","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.11613","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":null,"license_id":null,"version":null,"is_accepted":false,"is_published":null,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"doi:10.48550/arxiv.2605.11613","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.11613","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":null,"license_id":null,"version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"},"sustainable_development_goals":[{"display_name":"Quality Education","score":0.6637479066848755,"id":"https://metadata.un.org/sdg/4"}],"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],"self-distillation":[1,71],"has":[2],"emerged":[3],"as":[4,22],"a":[5,56,75,137,146,150],"promising":[6],"paradigm":[7],"for":[8,149],"post-training":[9],"language":[10],"models,":[11],"in":[12,63],"which":[13,131],"the":[14,64,70,84,89,92,95,113,117,133,141,176],"model":[15,172],"conditions":[16],"on":[17,121,170],"environment":[18],"feedback":[19,60,93],"to":[20],"serve":[21],"its":[23,38],"own":[24],"teacher,":[25],"providing":[26],"dense":[27],"token-level":[28],"rewards":[29],"without":[30],"external":[31],"teacher":[32,114],"models":[33],"or":[34,105],"step-level":[35],"annotations.":[36],"Despite":[37],"empirical":[39],"success,":[40],"what":[41,47],"this":[42,122],"reward":[43,73],"actually":[44],"measures":[45],"and":[46,91,167],"kind":[48],"of":[49,59],"credit":[50],"it":[51],"assigns":[52],"remain":[53],"unclear.":[54],"Under":[55],"posterior-compatibility":[57],"interpretation":[58],"conditioning,":[61],"standard":[62],"implicit-reward":[65],"literature,":[66],"we":[67,110,124],"show":[68],"that":[69,154],"token":[72],"is":[74,82,145],"Bayesian":[76],"filtering":[77],"increment":[78],"whose":[79],"trajectory":[80],"sum":[81],"exactly":[83],"pointwise":[85],"mutual":[86],"information":[87],"between":[88],"response":[90],"given":[94],"input.":[96],"This":[97],"pMI":[98,152],"can":[99],"be":[100],"raised":[101],"by":[102,106],"input-specific":[103,134],"reasoning":[104],"input-generic":[107],"shortcuts,":[108],"so":[109],"further":[111],"decompose":[112],"log-probability":[115],"along":[116],"input":[118],"axis.":[119],"Based":[120],"analysis,":[123],"propose":[125],"CREDIT":[126,144,174],"(Contrastive":[127],"REward":[128],"from":[129],"DIsTillation),":[130],"isolates":[132],"component":[135],"with":[136],"batch-contrastive":[138],"baseline.":[139],"At":[140],"sequence":[142],"level,":[143],"teacher-side":[147],"surrogate":[148],"contrastive":[151],"objective":[153],"also":[155],"penalizes":[156],"responses":[157],"remaining":[158],"likely":[159],"under":[160],"unrelated":[161],"inputs.":[162],"Across":[163],"coding,":[164],"scientific":[165],"reasoning,":[166],"tool-use":[168],"benchmarks":[169],"two":[171],"families,":[173],"delivers":[175],"strongest":[177],"aggregate":[178],"performance":[179],"at":[180],"negligible":[181],"additional":[182],"compute.":[183]},"counts_by_year":[],"updated_date":"2026-07-01T06:00:48.157686","created_date":"2026-05-14T00:00:00"}