{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2026:WE6ONQ4VWP4JBDXVC7YOYNV5SF","short_pith_number":"pith:WE6ONQ4V","canonical_record":{"source":{"id":"2605.12400","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-05-12T17:00:53Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"4496316516826098f6407c5ab502ca827ecb01db1d825e03d18de392fd5ba39f","abstract_canon_sha256":"a8c9c992aed92afea13155de0dc839558215e5b084a3f35b61b942ac9967fdda"},"schema_version":"1.0"},"canonical_sha256":"b13ce6c395b3f8908ef517f0ec36bd9178cc3a3844be93dfdc059ab8275b4f73","source":{"kind":"arxiv","id":"2605.12400","version":2},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.12400","created_at":"2026-06-02T01:03:48Z"},{"alias_kind":"arxiv_version","alias_value":"2605.12400v2","created_at":"2026-06-02T01:03:48Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.12400","created_at":"2026-06-02T01:03:48Z"},{"alias_kind":"pith_short_12","alias_value":"WE6ONQ4VWP4J","created_at":"2026-06-02T01:03:48Z"},{"alias_kind":"pith_short_16","alias_value":"WE6ONQ4VWP4JBDXV","created_at":"2026-06-02T01:03:48Z"},{"alias_kind":"pith_short_8","alias_value":"WE6ONQ4V","created_at":"2026-06-02T01:03:48Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2026:WE6ONQ4VWP4JBDXVC7YOYNV5SF","target":"record","payload":{"canonical_record":{"source":{"id":"2605.12400","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-05-12T17:00:53Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"4496316516826098f6407c5ab502ca827ecb01db1d825e03d18de392fd5ba39f","abstract_canon_sha256":"a8c9c992aed92afea13155de0dc839558215e5b084a3f35b61b942ac9967fdda"},"schema_version":"1.0"},"canonical_sha256":"b13ce6c395b3f8908ef517f0ec36bd9178cc3a3844be93dfdc059ab8275b4f73","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-02T01:03:48.438591Z","signature_b64":"qV1on+NiRFbGZvUYk437cpA2wmgN+5OUpowBpfvVVeSf0C2Hc5iYKHgP5y7E5jFkEujP45oWjvXluHvYDu+VDQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"b13ce6c395b3f8908ef517f0ec36bd9178cc3a3844be93dfdc059ab8275b4f73","last_reissued_at":"2026-06-02T01:03:48.438082Z","signature_status":"signed_v1","first_computed_at":"2026-06-02T01:03:48.438082Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2605.12400","source_version":2,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-06-02T01:03:48Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"b5uFP+Pd4ZW/MVVFTogR4+7tT+S7T75np02eer1y/Dqdb8NB+IY08Ra68FWdM78Mx1hmP+FD8tzYjLFklcMYCg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-06-12T04:00:20.568260Z"},"content_sha256":"0b18f2fa6ff4a429e3ca870dc41997ade3ec29ae9fce645a8af46eb59ab18dcf","schema_version":"1.0","event_id":"sha256:0b18f2fa6ff4a429e3ca870dc41997ade3ec29ae9fce645a8af46eb59ab18dcf"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2026:WE6ONQ4VWP4JBDXVC7YOYNV5SF","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"OGLS-SD: On-Policy Self-Distillation with Outcome-Guided Logit Steering for LLM Reasoning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"Outcome-guided logit steering calibrates teacher responses in on-policy self-distillation by contrasting successful and failed trajectories, reducing reflection bias for better LLM reasoning.","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Weitong Zhang, Xiaoyun Wang, Yuxiao Yang","submitted_at":"2026-05-12T17:00:53Z","abstract_excerpt":"We study on-policy self-distillation (OPSD), where a language model improves its reasoning ability by distilling privileged teacher distributions along its own on-policy trajectories. Despite its promise, OPSD can suffer from training instability due to a pattern mismatch between teacher and student responses. Self-reflected teacher responses may introduce reflection-induced biases and response templates that miscalibrate token-level supervision, ultimately harming the student's reasoning ability. To mitigate this issue, we propose OGLS-SD, an outcome-guided logit-steering framework that lever"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"By combining outcome-level correctness with dense token-level guidance through logit steering, OGLS-SD stabilizes self-distillation and improves reasoning performance over standard OPSD and other variants across diverse benchmarks.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That verifiable outcome rewards can reliably contrast successful and failed on-policy trajectories to calibrate teacher logits and mitigate reflection-induced bias without introducing new calibration issues or depending on tasks where outcomes are easily verifiable.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"OGLS-SD improves LLM reasoning by using verifiable outcome rewards to guide logit steering that calibrates teacher distributions in on-policy self-distillation, addressing reflection-induced mismatches.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Outcome-guided logit steering calibrates teacher responses in on-policy self-distillation by contrasting successful and failed trajectories, reducing reflection bias for better LLM reasoning.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"a34ed2143aab1182847d78b1fc606f13e67ee4443a9575f7535b78c8165d112d"},"source":{"id":"2605.12400","kind":"arxiv","version":2},"verdict":{"id":"3c816f6c-41af-4df3-8077-fa5a10a1946d","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-13T05:14:01.558810Z","strongest_claim":"By combining outcome-level correctness with dense token-level guidance through logit steering, OGLS-SD stabilizes self-distillation and improves reasoning performance over standard OPSD and other variants across diverse benchmarks.","one_line_summary":"OGLS-SD improves LLM reasoning by using verifiable outcome rewards to guide logit steering that calibrates teacher distributions in on-policy self-distillation, addressing reflection-induced mismatches.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That verifiable outcome rewards can reliably contrast successful and failed on-policy trajectories to calibrate teacher logits and mitigate reflection-induced bias without introducing new calibration issues or depending on tasks where outcomes are easily verifiable.","pith_extraction_headline":"Outcome-guided logit steering calibrates teacher responses in on-policy self-distillation by contrasting successful and failed trajectories, reducing reflection bias for better LLM reasoning."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2605.12400/integrity.json","findings":[],"available":true,"detectors_run":[{"name":"ai_meta_artifact","ran_at":"2026-05-26T14:40:00.497927Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"doi_title_agreement","ran_at":"2026-05-20T13:31:24.992235Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"doi_compliance","ran_at":"2026-05-20T09:37:46.618193Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"claim_evidence","ran_at":"2026-05-19T22:21:57.872449Z","status":"completed","version":"1.0.0","findings_count":0}],"snapshot_sha256":"4ccc60c6e5dbd90b21ad39be1748bbd4cf3e8a42ec904c34d6b7d211b50113cc"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":2,"snapshot_sha256":"bd398c76ef47e7aeaca8a0939b173d541fae73171e53f59e79da53c3bf777d43"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"verdict_id":"3c816f6c-41af-4df3-8077-fa5a10a1946d"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-06-02T01:03:48Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"RWj/SMYdYh+MmOPwCahka7j1J+h0FRKw7tI5smN9t226h5QxkTC8t3GAktMfcmmX9WfVoLGSZjfehVIyyz9pAw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-06-12T04:00:20.569450Z"},"content_sha256":"3798fbd20796eba654bd067e41f63c7048b3c164f714c5d0faababfffa2bb4bb","schema_version":"1.0","event_id":"sha256:3798fbd20796eba654bd067e41f63c7048b3c164f714c5d0faababfffa2bb4bb"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/WE6ONQ4VWP4JBDXVC7YOYNV5SF/bundle.json","state_url":"https://pith.science/pith/WE6ONQ4VWP4JBDXVC7YOYNV5SF/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/WE6ONQ4VWP4JBDXVC7YOYNV5SF/bundle.json","status":"primary"}],"public_keys":[{"key_id":"pith-v1-2026-05","algorithm":"ed25519","format":"raw","public_key_b64":"stVStoiQhXFxp4s2pdzPNoqVNBMojDU/fJ2db5S3CbM=","public_key_hex":"b2d552b68890857171a78b36a5dccf368a953413288c353f7c9d9d6f94b709b3","fingerprint_sha256_b32_first128bits":"RVFV5Z2OI2J3ZUO7ERDEBCYNKS","fingerprint_sha256_hex":"8d4b5ee74e4693bcd1df2446408b0d54","rotates_at":null,"url":"https://pith.science/pith-signing-key.json","notes":"Pith uses this Ed25519 key to sign canonical record SHA-256 digests. Verify with: ed25519_verify(public_key, message=canonical_sha256_bytes, signature=base64decode(signature_b64))."}],"merge_version":"pith-open-graph-merge-v1","built_at":"2026-06-12T04:00:20Z","links":{"resolver":"https://pith.science/pith/WE6ONQ4VWP4JBDXVC7YOYNV5SF","bundle":"https://pith.science/pith/WE6ONQ4VWP4JBDXVC7YOYNV5SF/bundle.json","state":"https://pith.science/pith/WE6ONQ4VWP4JBDXVC7YOYNV5SF/state.json","well_known_bundle":"https://pith.science/.well-known/pith/WE6ONQ4VWP4JBDXVC7YOYNV5SF/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:WE6ONQ4VWP4JBDXVC7YOYNV5SF","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":"a8c9c992aed92afea13155de0dc839558215e5b084a3f35b61b942ac9967fdda","cross_cats_sorted":["cs.AI"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-05-12T17:00:53Z","title_canon_sha256":"4496316516826098f6407c5ab502ca827ecb01db1d825e03d18de392fd5ba39f"},"schema_version":"1.0","source":{"id":"2605.12400","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.12400","created_at":"2026-06-02T01:03:48Z"},{"alias_kind":"arxiv_version","alias_value":"2605.12400v2","created_at":"2026-06-02T01:03:48Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.12400","created_at":"2026-06-02T01:03:48Z"},{"alias_kind":"pith_short_12","alias_value":"WE6ONQ4VWP4J","created_at":"2026-06-02T01:03:48Z"},{"alias_kind":"pith_short_16","alias_value":"WE6ONQ4VWP4JBDXV","created_at":"2026-06-02T01:03:48Z"},{"alias_kind":"pith_short_8","alias_value":"WE6ONQ4V","created_at":"2026-06-02T01:03:48Z"}],"graph_snapshots":[{"event_id":"sha256:3798fbd20796eba654bd067e41f63c7048b3c164f714c5d0faababfffa2bb4bb","target":"graph","created_at":"2026-06-02T01:03:48Z","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":"By combining outcome-level correctness with dense token-level guidance through logit steering, OGLS-SD stabilizes self-distillation and improves reasoning performance over standard OPSD and other variants across diverse benchmarks."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"That verifiable outcome rewards can reliably contrast successful and failed on-policy trajectories to calibrate teacher logits and mitigate reflection-induced bias without introducing new calibration issues or depending on tasks where outcomes are easily verifiable."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"OGLS-SD improves LLM reasoning by using verifiable outcome rewards to guide logit steering that calibrates teacher distributions in on-policy self-distillation, addressing reflection-induced mismatches."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"Outcome-guided logit steering calibrates teacher responses in on-policy self-distillation by contrasting successful and failed trajectories, reducing reflection bias for better LLM reasoning."}],"snapshot_sha256":"a34ed2143aab1182847d78b1fc606f13e67ee4443a9575f7535b78c8165d112d"},"formal_canon":{"evidence_count":2,"snapshot_sha256":"bd398c76ef47e7aeaca8a0939b173d541fae73171e53f59e79da53c3bf777d43"},"integrity":{"available":true,"clean":true,"detectors_run":[{"findings_count":0,"name":"ai_meta_artifact","ran_at":"2026-05-26T14:40:00.497927Z","status":"completed","version":"1.0.0"},{"findings_count":0,"name":"doi_title_agreement","ran_at":"2026-05-20T13:31:24.992235Z","status":"completed","version":"1.0.0"},{"findings_count":0,"name":"doi_compliance","ran_at":"2026-05-20T09:37:46.618193Z","status":"completed","version":"1.0.0"},{"findings_count":0,"name":"claim_evidence","ran_at":"2026-05-19T22:21:57.872449Z","status":"completed","version":"1.0.0"}],"endpoint":"/pith/2605.12400/integrity.json","findings":[],"snapshot_sha256":"4ccc60c6e5dbd90b21ad39be1748bbd4cf3e8a42ec904c34d6b7d211b50113cc","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"We study on-policy self-distillation (OPSD), where a language model improves its reasoning ability by distilling privileged teacher distributions along its own on-policy trajectories. Despite its promise, OPSD can suffer from training instability due to a pattern mismatch between teacher and student responses. Self-reflected teacher responses may introduce reflection-induced biases and response templates that miscalibrate token-level supervision, ultimately harming the student's reasoning ability. To mitigate this issue, we propose OGLS-SD, an outcome-guided logit-steering framework that lever","authors_text":"Weitong Zhang, Xiaoyun Wang, Yuxiao Yang","cross_cats":["cs.AI"],"headline":"Outcome-guided logit steering calibrates teacher responses in on-policy self-distillation by contrasting successful and failed trajectories, reducing reflection bias for better LLM reasoning.","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-05-12T17:00:53Z","title":"OGLS-SD: On-Policy Self-Distillation with Outcome-Guided Logit Steering for LLM Reasoning"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.12400","kind":"arxiv","version":2},"verdict":{"created_at":"2026-05-13T05:14:01.558810Z","id":"3c816f6c-41af-4df3-8077-fa5a10a1946d","model_set":{"reader":"grok-4.3"},"one_line_summary":"OGLS-SD improves LLM reasoning by using verifiable outcome rewards to guide logit steering that calibrates teacher distributions in on-policy self-distillation, addressing reflection-induced mismatches.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"Outcome-guided logit steering calibrates teacher responses in on-policy self-distillation by contrasting successful and failed trajectories, reducing reflection bias for better LLM reasoning.","strongest_claim":"By combining outcome-level correctness with dense token-level guidance through logit steering, OGLS-SD stabilizes self-distillation and improves reasoning performance over standard OPSD and other variants across diverse benchmarks.","weakest_assumption":"That verifiable outcome rewards can reliably contrast successful and failed on-policy trajectories to calibrate teacher logits and mitigate reflection-induced bias without introducing new calibration issues or depending on tasks where outcomes are easily verifiable."}},"verdict_id":"3c816f6c-41af-4df3-8077-fa5a10a1946d"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:0b18f2fa6ff4a429e3ca870dc41997ade3ec29ae9fce645a8af46eb59ab18dcf","target":"record","created_at":"2026-06-02T01:03:48Z","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":"a8c9c992aed92afea13155de0dc839558215e5b084a3f35b61b942ac9967fdda","cross_cats_sorted":["cs.AI"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-05-12T17:00:53Z","title_canon_sha256":"4496316516826098f6407c5ab502ca827ecb01db1d825e03d18de392fd5ba39f"},"schema_version":"1.0","source":{"id":"2605.12400","kind":"arxiv","version":2}},"canonical_sha256":"b13ce6c395b3f8908ef517f0ec36bd9178cc3a3844be93dfdc059ab8275b4f73","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"b13ce6c395b3f8908ef517f0ec36bd9178cc3a3844be93dfdc059ab8275b4f73","first_computed_at":"2026-06-02T01:03:48.438082Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-06-02T01:03:48.438082Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"qV1on+NiRFbGZvUYk437cpA2wmgN+5OUpowBpfvVVeSf0C2Hc5iYKHgP5y7E5jFkEujP45oWjvXluHvYDu+VDQ==","signature_status":"signed_v1","signed_at":"2026-06-02T01:03:48.438591Z","signed_message":"canonical_sha256_bytes"},"source_id":"2605.12400","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:0b18f2fa6ff4a429e3ca870dc41997ade3ec29ae9fce645a8af46eb59ab18dcf","sha256:3798fbd20796eba654bd067e41f63c7048b3c164f714c5d0faababfffa2bb4bb"],"state_sha256":"fa817e27fcdbe17aadcd79c2af95463066a54d6b9ff332e5d47b301ef4fd85bc"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"BTiGD/foeVAR/XBtjxbK2w3nX3JxUSHQKMU+ocEcPN3PysnBqAzZkSl88yrLHZ1zOn4AgnzLLLFyrF6rvzmsBw==","signed_message":"bundle_sha256_bytes","signed_at":"2026-06-12T04:00:20.575863Z","bundle_sha256":"458580a27aa85dbab57a7f50457d8b111914c14fe1d666af22d258516c6bc42b"}}