{"id":"https://openalex.org/W7164539611","doi":"https://doi.org/10.48550/arxiv.2606.12507","title":"Rubric-Guided Self-Distillation: Post-Training Without Rubric Verifiers","display_name":"Rubric-Guided Self-Distillation: Post-Training Without Rubric Verifiers","publication_year":2026,"publication_date":"2026-06-10","ids":{"openalex":"https://openalex.org/W7164539611","doi":"https://doi.org/10.48550/arxiv.2606.12507"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2606.12507","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.12507","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.2606.12507","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5138530580","display_name":"MohammadHossein Rezaei","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Rezaei, MohammadHossein","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5136018321","display_name":"Anas Mahmoud","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Mahmoud, Anas","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5138527160","display_name":"Zihao Wang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wang, Zihao","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5138547193","display_name":"Utkarsh Tyagi","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Tyagi, Utkarsh","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5138512476","display_name":"Advait Gosai","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Gosai, Advait","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5036739449","display_name":"Razvan-Gabriel Dumitru","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Dumitru, Razvan-Gabriel","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5090283848","display_name":"Aakash Sabharwal","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Sabharwal, Aakash","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5138496592","display_name":"Bing Liu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Liu, Bing","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5121999607","display_name":"Y He","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"He, Yunzhong","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":0.19380000233650208,"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":0.19380000233650208,"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":0.1185000017285347,"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/T11689","display_name":"Adversarial Robustness in Machine Learning","score":0.08470000326633453,"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":0.9797000288963318},{"id":"https://openalex.org/keywords/reliability","display_name":"Reliability (semiconductor)","score":0.5634999871253967},{"id":"https://openalex.org/keywords/class","display_name":"Class (philosophy)","score":0.322299987077713},{"id":"https://openalex.org/keywords/raw-data","display_name":"Raw data","score":0.3206999897956848},{"id":"https://openalex.org/keywords/signal","display_name":"SIGNAL (programming language)","score":0.3203999996185303}],"concepts":[{"id":"https://openalex.org/C111640148","wikidata":"https://www.wikidata.org/wiki/Q847349","display_name":"Rubric","level":2,"score":0.9797000288963318},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5920000076293945},{"id":"https://openalex.org/C43214815","wikidata":"https://www.wikidata.org/wiki/Q7310987","display_name":"Reliability (semiconductor)","level":3,"score":0.5634999871253967},{"id":"https://openalex.org/C145420912","wikidata":"https://www.wikidata.org/wiki/Q853077","display_name":"Mathematics education","level":1,"score":0.4056999981403351},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.35370001196861267},{"id":"https://openalex.org/C2777212361","wikidata":"https://www.wikidata.org/wiki/Q5127848","display_name":"Class (philosophy)","level":2,"score":0.322299987077713},{"id":"https://openalex.org/C132964779","wikidata":"https://www.wikidata.org/wiki/Q2110223","display_name":"Raw data","level":2,"score":0.3206999897956848},{"id":"https://openalex.org/C2779843651","wikidata":"https://www.wikidata.org/wiki/Q7390335","display_name":"SIGNAL (programming language)","level":2,"score":0.3203999996185303},{"id":"https://openalex.org/C15744967","wikidata":"https://www.wikidata.org/wiki/Q9418","display_name":"Psychology","level":0,"score":0.3111000061035156},{"id":"https://openalex.org/C42058472","wikidata":"https://www.wikidata.org/wiki/Q810214","display_name":"Base (topology)","level":2,"score":0.3075999915599823},{"id":"https://openalex.org/C51632099","wikidata":"https://www.wikidata.org/wiki/Q3985153","display_name":"Training set","level":2,"score":0.30550000071525574},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.2921999990940094}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2606.12507","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.12507","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.2606.12507","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.12507","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":0.5953168869018555,"id":"https://metadata.un.org/sdg/4","display_name":"Quality Education"}],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Rubrics":[0],"have":[1],"emerged":[2],"as":[3,73,171],"an":[4,26],"alternative":[5,175],"to":[6,42,49,128],"RLVR":[7],"in":[8,63,166],"open-ended":[9],"domains":[10],"where":[11],"a":[12,50,59,149,159,172],"single":[13],"ground-truth":[14],"final":[15],"answer":[16],"is":[17,181],"not":[18],"available.":[19],"Existing":[20],"rubric-based":[21],"training":[22,61,106],"methods":[23],"rely":[24],"on":[25,69,118],"LLM":[27,102],"verifier":[28,141,177],"that":[29,145],"scores":[30],"each":[31],"rollout":[32,135],"against":[33],"rubrics.":[34],"This":[35],"introduces":[36],"substantial":[37],"training-time":[38,140],"overhead,":[39],"exposes":[40],"optimization":[41],"verifier-specific":[43],"biases,":[44],"and":[45,99,113,120,138],"reduces":[46],"rubric":[47,125],"feedback":[48],"sparse":[51,91],"end-of-trajectory":[52],"signal.":[53],"We":[54],"propose":[55],"Rubric-Guided":[56],"Self-Distillation":[57],"(RGSD),":[58],"verifier-free":[60,174],"method":[62],"which":[64],"the":[65,70,74,77,82,87,101,105,182],"base":[66],"policy,":[67],"conditioned":[68],"rubric,":[71],"serves":[72],"teacher":[75,84,151],"for":[76],"unconditioned":[78],"student.":[79],"RGSD":[80,123,165,170],"distills":[81],"rubric-conditioned":[83],"distribution":[85],"into":[86],"student":[88],"token-by-token,":[89],"replacing":[90],"trajectory-level":[92],"rewards":[93],"with":[94],"dense":[95],"per-token":[96],"learning":[97],"signals":[98],"removing":[100],"judge":[103,162],"from":[104],"loop":[107],"entirely.":[108],"Across":[109],"Qwen-2.5":[110],"(3B,":[111],"7B)":[112],"Qwen3-Thinking":[114],"(4B,":[115],"8B)":[116],"models":[117],"medical":[119],"science":[121],"domains,":[122],"achieves":[124],"satisfaction":[126],"comparable":[127],"judge-based":[129],"GRPO":[130,161],"while":[131,158],"using":[132],"one":[133],"on-policy":[134],"per":[136],"prompt":[137],"no":[139],"calls.":[142],"Ablations":[143],"show":[144],"raw":[146],"rubrics":[147],"provide":[148],"stronger":[150,160],"enrichment":[152],"signal":[153],"than":[154],"self-generated":[155],"reference":[156],"responses,":[157],"can":[163],"outperform":[164],"some":[167],"settings,":[168],"positioning":[169],"complementary":[173],"when":[176],"cost":[178],"or":[179],"reliability":[180],"bottleneck.":[183]},"counts_by_year":[],"updated_date":"2026-07-28T07:46:37.118299","created_date":"2026-06-13T00:00:00"}