{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:HBBWPG6DRQYAPR7CAMCKSJHRA4","short_pith_number":"pith:HBBWPG6D","schema_version":"1.0","canonical_sha256":"3843679bc38c3007c7e20304a924f1073c65710c41ea8c89574dd0d2d38fc2c2","source":{"kind":"arxiv","id":"2606.07006","version":1},"attestation_state":"computed","paper":{"title":"RASFT: Rollout-Adaptive Supervised Fine-Tuning for Reasoning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.LG","authors_text":"Fei Sun, Fengyuan Liu, Mengnan Du, Na Zou, Wei Shi, Yanguang Liu, Yongliang Miao","submitted_at":"2026-06-05T07:52:40Z","abstract_excerpt":"Supervised fine-tuning (SFT) is a prevailing method for adapting large language models to reasoning tasks by imitating offline expert demonstrations, often treating a single expert trajectory as the target behavior. However, reasoning is not simple path imitation: rigidly following one demonstrated solution may overfit to surface forms and suppress the model's own reasoning distribution. We propose Rollout-Adaptive Supervised Fine-Tuning (RASFT), a policy-aware SFT framework that calibrates expert supervision according to problem-level solvability estimated from verified on-policy rollouts. Fo"},"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":"2606.07006","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-06-05T07:52:40Z","cross_cats_sorted":["cs.CL"],"title_canon_sha256":"27e00b93e8d485d3e6f838496472ea78aa3a0cb4b9514e526a4319445ce75cd2","abstract_canon_sha256":"3607e5e98d8bec6ce90a5a8699454013ea83146fb5381426eadbb64488046564"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-08T01:04:40.840677Z","signature_b64":"lG11HvcjPwVfLE2JQzT5ZugEYYWW7eCrIbCwsSNiAxAU0HOUbkIvBAPdbbX3IjWu+Jfgs71dljM1DAxyfB6TAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"3843679bc38c3007c7e20304a924f1073c65710c41ea8c89574dd0d2d38fc2c2","last_reissued_at":"2026-06-08T01:04:40.839829Z","signature_status":"signed_v1","first_computed_at":"2026-06-08T01:04:40.839829Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"RASFT: Rollout-Adaptive Supervised Fine-Tuning for Reasoning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.LG","authors_text":"Fei Sun, Fengyuan Liu, Mengnan Du, Na Zou, Wei Shi, Yanguang Liu, Yongliang Miao","submitted_at":"2026-06-05T07:52:40Z","abstract_excerpt":"Supervised fine-tuning (SFT) is a prevailing method for adapting large language models to reasoning tasks by imitating offline expert demonstrations, often treating a single expert trajectory as the target behavior. However, reasoning is not simple path imitation: rigidly following one demonstrated solution may overfit to surface forms and suppress the model's own reasoning distribution. We propose Rollout-Adaptive Supervised Fine-Tuning (RASFT), a policy-aware SFT framework that calibrates expert supervision according to problem-level solvability estimated from verified on-policy rollouts. Fo"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.07006","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/2606.07006/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":"2606.07006","created_at":"2026-06-08T01:04:40.839916+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.07006v1","created_at":"2026-06-08T01:04:40.839916+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.07006","created_at":"2026-06-08T01:04:40.839916+00:00"},{"alias_kind":"pith_short_12","alias_value":"HBBWPG6DRQYA","created_at":"2026-06-08T01:04:40.839916+00:00"},{"alias_kind":"pith_short_16","alias_value":"HBBWPG6DRQYAPR7C","created_at":"2026-06-08T01:04:40.839916+00:00"},{"alias_kind":"pith_short_8","alias_value":"HBBWPG6D","created_at":"2026-06-08T01:04:40.839916+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/HBBWPG6DRQYAPR7CAMCKSJHRA4","json":"https://pith.science/pith/HBBWPG6DRQYAPR7CAMCKSJHRA4.json","graph_json":"https://pith.science/api/pith-number/HBBWPG6DRQYAPR7CAMCKSJHRA4/graph.json","events_json":"https://pith.science/api/pith-number/HBBWPG6DRQYAPR7CAMCKSJHRA4/events.json","paper":"https://pith.science/paper/HBBWPG6D"},"agent_actions":{"view_html":"https://pith.science/pith/HBBWPG6DRQYAPR7CAMCKSJHRA4","download_json":"https://pith.science/pith/HBBWPG6DRQYAPR7CAMCKSJHRA4.json","view_paper":"https://pith.science/paper/HBBWPG6D","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.07006&json=true","fetch_graph":"https://pith.science/api/pith-number/HBBWPG6DRQYAPR7CAMCKSJHRA4/graph.json","fetch_events":"https://pith.science/api/pith-number/HBBWPG6DRQYAPR7CAMCKSJHRA4/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/HBBWPG6DRQYAPR7CAMCKSJHRA4/action/timestamp_anchor","attest_storage":"https://pith.science/pith/HBBWPG6DRQYAPR7CAMCKSJHRA4/action/storage_attestation","attest_author":"https://pith.science/pith/HBBWPG6DRQYAPR7CAMCKSJHRA4/action/author_attestation","sign_citation":"https://pith.science/pith/HBBWPG6DRQYAPR7CAMCKSJHRA4/action/citation_signature","submit_replication":"https://pith.science/pith/HBBWPG6DRQYAPR7CAMCKSJHRA4/action/replication_record"}},"created_at":"2026-06-08T01:04:40.839916+00:00","updated_at":"2026-06-08T01:04:40.839916+00:00"}