{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:LP2MQEYNGEGACGEBS3TYWG5UY2","short_pith_number":"pith:LP2MQEYN","schema_version":"1.0","canonical_sha256":"5bf4c8130d310c01188196e78b1bb4c6b10b381b1bbfda3a90c3f6eec64b6d24","source":{"kind":"arxiv","id":"2607.18955","version":1},"attestation_state":"computed","paper":{"title":"H$^2$SD: Hybrid Hindsight Self-Distillation","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.LG","authors_text":"Bing Qin, Linyang Li, Peiji Li, Qipeng Guo, Qiye Cai, Tao Gui, Xiaocheng Feng, Yicheng Zou, Yichuan Ma, Yongkang Chen","submitted_at":"2026-07-21T10:47:27Z","abstract_excerpt":"Reinforcement learning with verifiable rewards (RLVR) has substantially improved the reasoning capabilities of large language models on tasks such as mathematical reasoning and code generation. However, most RLVR methods assign a scalar outcome reward to an entire trajectory, resulting in sparse supervision and limited token-level credit assignment. On-policy distillation (OPD) provides denser supervision by distilling token-level distributions from a stronger teacher model, but requires an additional teacher and typically assumes a shared vocabulary. On-policy self-distillation (OPSD) removes"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2607.18955","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-07-21T10:47:27Z","cross_cats_sorted":["cs.CL"],"title_canon_sha256":"0cbaedbc4229e95ae9ae59d72b344963c09f9aeaf0e4f47656547dea7524f285","abstract_canon_sha256":"72f2949291cfa6899cc70238d10f9e68d0ea5a5f2126bd07017398f5bcdf46c7"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-22T01:23:21.661741Z","signature_b64":"WwISb91sz68cfb1ucawZnmFgomQmZyUL0Pxxrgh+AT7gGL2IUBhWaHl7Z6KZqeUargaH/4kg4cfOWzDgy5GJDQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"5bf4c8130d310c01188196e78b1bb4c6b10b381b1bbfda3a90c3f6eec64b6d24","last_reissued_at":"2026-07-22T01:23:21.660940Z","signature_status":"signed_v1","first_computed_at":"2026-07-22T01:23:21.660940Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"H$^2$SD: Hybrid Hindsight Self-Distillation","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.LG","authors_text":"Bing Qin, Linyang Li, Peiji Li, Qipeng Guo, Qiye Cai, Tao Gui, Xiaocheng Feng, Yicheng Zou, Yichuan Ma, Yongkang Chen","submitted_at":"2026-07-21T10:47:27Z","abstract_excerpt":"Reinforcement learning with verifiable rewards (RLVR) has substantially improved the reasoning capabilities of large language models on tasks such as mathematical reasoning and code generation. However, most RLVR methods assign a scalar outcome reward to an entire trajectory, resulting in sparse supervision and limited token-level credit assignment. On-policy distillation (OPD) provides denser supervision by distilling token-level distributions from a stronger teacher model, but requires an additional teacher and typically assumes a shared vocabulary. On-policy self-distillation (OPSD) removes"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2607.18955","kind":"arxiv","version":1},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2607.18955/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"aliases":[{"alias_kind":"arxiv","alias_value":"2607.18955","created_at":"2026-07-22T01:23:21.661337+00:00"},{"alias_kind":"arxiv_version","alias_value":"2607.18955v1","created_at":"2026-07-22T01:23:21.661337+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2607.18955","created_at":"2026-07-22T01:23:21.661337+00:00"},{"alias_kind":"pith_short_12","alias_value":"LP2MQEYNGEGA","created_at":"2026-07-22T01:23:21.661337+00:00"},{"alias_kind":"pith_short_16","alias_value":"LP2MQEYNGEGACGEB","created_at":"2026-07-22T01:23:21.661337+00:00"},{"alias_kind":"pith_short_8","alias_value":"LP2MQEYN","created_at":"2026-07-22T01:23:21.661337+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/LP2MQEYNGEGACGEBS3TYWG5UY2","json":"https://pith.science/pith/LP2MQEYNGEGACGEBS3TYWG5UY2.json","graph_json":"https://pith.science/api/pith-number/LP2MQEYNGEGACGEBS3TYWG5UY2/graph.json","events_json":"https://pith.science/api/pith-number/LP2MQEYNGEGACGEBS3TYWG5UY2/events.json","paper":"https://pith.science/paper/LP2MQEYN"},"agent_actions":{"view_html":"https://pith.science/pith/LP2MQEYNGEGACGEBS3TYWG5UY2","download_json":"https://pith.science/pith/LP2MQEYNGEGACGEBS3TYWG5UY2.json","view_paper":"https://pith.science/paper/LP2MQEYN","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2607.18955&json=true","fetch_graph":"https://pith.science/api/pith-number/LP2MQEYNGEGACGEBS3TYWG5UY2/graph.json","fetch_events":"https://pith.science/api/pith-number/LP2MQEYNGEGACGEBS3TYWG5UY2/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/LP2MQEYNGEGACGEBS3TYWG5UY2/action/timestamp_anchor","attest_storage":"https://pith.science/pith/LP2MQEYNGEGACGEBS3TYWG5UY2/action/storage_attestation","attest_author":"https://pith.science/pith/LP2MQEYNGEGACGEBS3TYWG5UY2/action/author_attestation","sign_citation":"https://pith.science/pith/LP2MQEYNGEGACGEBS3TYWG5UY2/action/citation_signature","submit_replication":"https://pith.science/pith/LP2MQEYNGEGACGEBS3TYWG5UY2/action/replication_record"}},"created_at":"2026-07-22T01:23:21.661337+00:00","updated_at":"2026-07-22T01:23:21.661337+00:00"}