{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:XP5QRJWORXVGYL5LSNZ3LPVQYR","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"a2a6d74918e4bf28b16bc80f1d5e27a016ac55827482ebac9d171647432ca76e","cross_cats_sorted":["cs.AI"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-10-21T17:15:56Z","title_canon_sha256":"3da41121e66a06f632b807a7bacb8a67fc88c090f8e70453dc0c1151ff1a4d99"},"schema_version":"1.0","source":{"id":"2510.18814","kind":"arxiv","version":3}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2510.18814","created_at":"2026-05-20T00:00:26Z"},{"alias_kind":"arxiv_version","alias_value":"2510.18814v3","created_at":"2026-05-20T00:00:26Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2510.18814","created_at":"2026-05-20T00:00:26Z"},{"alias_kind":"pith_short_12","alias_value":"XP5QRJWORXVG","created_at":"2026-05-20T00:00:26Z"},{"alias_kind":"pith_short_16","alias_value":"XP5QRJWORXVGYL5L","created_at":"2026-05-20T00:00:26Z"},{"alias_kind":"pith_short_8","alias_value":"XP5QRJWO","created_at":"2026-05-20T00:00:26Z"}],"graph_snapshots":[{"event_id":"sha256:0dc05f0ea663182f3d4434db1b623211653608d804447866a4be346ecfaffc66","target":"graph","created_at":"2026-05-20T00:00:26Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":4,"items":[{"attestation":"unclaimed","claim_id":"C1","kind":"strongest_claim","source":"verdict.strongest_claim","status":"machine_extracted","text":"We show that they can. We propose Self-evolving Post-Training (SePT), a simple post-training method that alternates between self-generation and training on self-generated responses. Across six math reasoning benchmarks, SePT improves a strong no-training baseline... and in some settings can even approach the performance of Reinforcement Learning with Verifiable Rewards (RLVR)."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"That self-generated responses supply a net positive training signal rather than reinforcing the model's existing errors or hallucinations, which must hold for the iterative self-training loop to produce sustained gains without external verification or filtering."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"SePT enables LLMs to improve math reasoning on multiple benchmarks by iteratively training on their own low-temperature generated responses using an online data refresh mechanism."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"Language models can improve their reasoning by training on responses they generate themselves."}],"snapshot_sha256":"d1eefac338f9d4b3286071d0d1691c85c584bb514eff26205aa22a72cc14b1da"},"formal_canon":{"evidence_count":2,"snapshot_sha256":"1da77da5bf2e75e62ba903468a8c56f50941e63395d2864ab68fa7a509218b65"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/2510.18814/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Can language models improve their reasoning performance without external rewards, using only their own sampled responses for training? We show that they can. We propose Self-evolving Post-Training (SePT), a simple post-training method that alternates between self-generation and training on self-generated responses. It repeatedly samples questions, uses the model itself to generate responses under a specified sampling temperature, and then trains the model on the self-generated data. In this self-training loop, we use an online data refresh mechanism, where each new batch is generated by the mo","authors_text":"Anthony Man-Cho So, Lei Zhao, Mengqi Li, Ruoyu Sun, Xiao Li","cross_cats":["cs.AI"],"headline":"Language models can improve their reasoning by training on responses they generate themselves.","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-10-21T17:15:56Z","title":"A Model Can Help Itself: Reward-Free Self-Training for LLM Reasoning"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2510.18814","kind":"arxiv","version":3},"verdict":{"created_at":"2026-05-18T05:07:29.377760Z","id":"f4d25a43-0b64-42de-9fb7-be4d16c84530","model_set":{"reader":"grok-4.3"},"one_line_summary":"SePT enables LLMs to improve math reasoning on multiple benchmarks by iteratively training on their own low-temperature generated responses using an online data refresh mechanism.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"Language models can improve their reasoning by training on responses they generate themselves.","strongest_claim":"We show that they can. We propose Self-evolving Post-Training (SePT), a simple post-training method that alternates between self-generation and training on self-generated responses. Across six math reasoning benchmarks, SePT improves a strong no-training baseline... and in some settings can even approach the performance of Reinforcement Learning with Verifiable Rewards (RLVR).","weakest_assumption":"That self-generated responses supply a net positive training signal rather than reinforcing the model's existing errors or hallucinations, which must hold for the iterative self-training loop to produce sustained gains without external verification or filtering."}},"verdict_id":"f4d25a43-0b64-42de-9fb7-be4d16c84530"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:ab51499825b1a9a7effa9637d296d2b76910b78742e74a11033178892db5a87a","target":"record","created_at":"2026-05-20T00:00:26Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"a2a6d74918e4bf28b16bc80f1d5e27a016ac55827482ebac9d171647432ca76e","cross_cats_sorted":["cs.AI"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-10-21T17:15:56Z","title_canon_sha256":"3da41121e66a06f632b807a7bacb8a67fc88c090f8e70453dc0c1151ff1a4d99"},"schema_version":"1.0","source":{"id":"2510.18814","kind":"arxiv","version":3}},"canonical_sha256":"bbfb08a6ce8dea6c2fab9373b5beb0c471ac5638b82199949a8de7ac157dbfa3","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"bbfb08a6ce8dea6c2fab9373b5beb0c471ac5638b82199949a8de7ac157dbfa3","first_computed_at":"2026-05-20T00:00:26.365399Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-20T00:00:26.365399Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"cwC3wfbzGfGfP6qgYXAuD2g7qmyeNg1/Qwk/Q7+CEi9L4hYBVxwID3tCQEq0xS12r8pcABoGihILaQsfa1mBAg==","signature_status":"signed_v1","signed_at":"2026-05-20T00:00:26.366372Z","signed_message":"canonical_sha256_bytes"},"source_id":"2510.18814","source_kind":"arxiv","source_version":3}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:ab51499825b1a9a7effa9637d296d2b76910b78742e74a11033178892db5a87a","sha256:0dc05f0ea663182f3d4434db1b623211653608d804447866a4be346ecfaffc66"],"state_sha256":"0349be25ee1126133481ccf4ae1098a4a279d3cf936b702e929d4b006270e683"}