{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2026:7XEA25FQG2QRI2FNXXF7MDASVD","short_pith_number":"pith:7XEA25FQ","canonical_record":{"source":{"id":"2605.06597","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.CL","submitted_at":"2026-05-07T17:22:11Z","cross_cats_sorted":["cs.AI","cs.LG"],"title_canon_sha256":"9ee1af973d04efab55c7bc8c341184da8f541824251ef46afd93b72e02342787","abstract_canon_sha256":"609fb5e8e5266b7b8008b4f1e30100c9dc1b5b20f5eb31361abcd749c2054462"},"schema_version":"1.0"},"canonical_sha256":"fdc80d74b036a11468adbdcbf60c12a8fba5c294aa7459c60fd9f78cb640d93d","source":{"kind":"arxiv","id":"2605.06597","version":2},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.06597","created_at":"2026-05-22T02:04:41Z"},{"alias_kind":"arxiv_version","alias_value":"2605.06597v2","created_at":"2026-05-22T02:04:41Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.06597","created_at":"2026-05-22T02:04:41Z"},{"alias_kind":"pith_short_12","alias_value":"7XEA25FQG2QR","created_at":"2026-05-22T02:04:41Z"},{"alias_kind":"pith_short_16","alias_value":"7XEA25FQG2QRI2FN","created_at":"2026-05-22T02:04:41Z"},{"alias_kind":"pith_short_8","alias_value":"7XEA25FQ","created_at":"2026-05-22T02:04:41Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2026:7XEA25FQG2QRI2FNXXF7MDASVD","target":"record","payload":{"canonical_record":{"source":{"id":"2605.06597","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.CL","submitted_at":"2026-05-07T17:22:11Z","cross_cats_sorted":["cs.AI","cs.LG"],"title_canon_sha256":"9ee1af973d04efab55c7bc8c341184da8f541824251ef46afd93b72e02342787","abstract_canon_sha256":"609fb5e8e5266b7b8008b4f1e30100c9dc1b5b20f5eb31361abcd749c2054462"},"schema_version":"1.0"},"canonical_sha256":"fdc80d74b036a11468adbdcbf60c12a8fba5c294aa7459c60fd9f78cb640d93d","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-22T02:04:41.945423Z","signature_b64":"gyF1yaXC3sPgmGSU6uyD6hY02kKytmu0mpyGTpZF+34YCgIDm6RgqMPnvAyCYwwIitOM/NliQncLk8k2SjIBCQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"fdc80d74b036a11468adbdcbf60c12a8fba5c294aa7459c60fd9f78cb640d93d","last_reissued_at":"2026-05-22T02:04:41.944621Z","signature_status":"signed_v1","first_computed_at":"2026-05-22T02:04:41.944621Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2605.06597","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":"2026-05-22T02:04:41Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"7RPjT/A0hMToeQHFy5Q59jnqt4juMZkGjGI6EAd019ai3w6P/lJ9KfFl79LVdS2/yP5wYON8Gc4ZRqDomSmNAw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-23T12:10:40.739594Z"},"content_sha256":"c659f9c633dfbdf3d883114d5456ce7d2121bc5bda13650907229d2cdfebf559","schema_version":"1.0","event_id":"sha256:c659f9c633dfbdf3d883114d5456ce7d2121bc5bda13650907229d2cdfebf559"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2026:7XEA25FQG2QRI2FNXXF7MDASVD","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"UniSD: Towards a Unified Self-Distillation Framework for Large Language Models","license":"http://creativecommons.org/licenses/by-sa/4.0/","headline":"A unified framework makes self-distillation a reliable way to adapt large language models without stronger teachers.","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"B. Aditya Prakash, Haoxin Liu, Jindong Wang, Josiah Hester, Lucheng Fu, Srijan Kumar, Yijia Xiao, Yinyi Luo, Yiqiao Jin, Yiyang Wang","submitted_at":"2026-05-07T17:22:11Z","abstract_excerpt":"Self-distillation (SD) offers a promising path for adapting large language models (LLMs) without relying on stronger external teachers. However, SD in autoregressive LLMs remains challenging because self-generated trajectories are free-form, correctness is task-dependent, and plausible rationales can still provide unstable or unreliable supervision. Existing methods mainly examine isolated design choices, leaving their effectiveness, roles, and interactions unclear. In this paper, we propose UniSD, a unified framework to systematically study self-distillation. UniSD integrates complementary me"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"UniSDfull, an integrated pipeline that combines complementary components, achieves the strongest overall performance, improving over the base model by +5.4 points and the strongest baseline by +2.8 points.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That the listed mechanisms (multi-teacher agreement, EMA, token contrastive learning, feature matching, divergence clipping) reliably address supervision instability in free-form self-generated trajectories and that their interactions can be isolated and combined without post-hoc selection bias affecting the reported gains.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"UniSD unifies complementary self-distillation mechanisms for autoregressive LLMs and achieves up to +5.4 point gains over base models and +2.8 over baselines across six benchmarks and six models.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"A unified framework makes self-distillation a reliable way to adapt large language models without stronger teachers.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"43f0993ffab088ed0c3a24d3818d0127f6c5b5d58b2ff8847fd9c854ec42b87a"},"source":{"id":"2605.06597","kind":"arxiv","version":2},"verdict":{"id":"c6a1c7da-9796-4d61-893c-c22d84397521","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-08T09:52:45.380637Z","strongest_claim":"UniSDfull, an integrated pipeline that combines complementary components, achieves the strongest overall performance, improving over the base model by +5.4 points and the strongest baseline by +2.8 points.","one_line_summary":"UniSD unifies complementary self-distillation mechanisms for autoregressive LLMs and achieves up to +5.4 point gains over base models and +2.8 over baselines across six benchmarks and six models.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That the listed mechanisms (multi-teacher agreement, EMA, token contrastive learning, feature matching, divergence clipping) reliably address supervision instability in free-form self-generated trajectories and that their interactions can be isolated and combined without post-hoc selection bias affecting the reported gains.","pith_extraction_headline":"A unified framework makes self-distillation a reliable way to adapt large language models without stronger teachers."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2605.06597/integrity.json","findings":[],"available":true,"detectors_run":[{"name":"claim_evidence","ran_at":"2026-05-20T12:22:03.785641Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"ai_meta_artifact","ran_at":"2026-05-20T07:37:52.552364Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"doi_title_agreement","ran_at":"2026-05-19T18:01:19.364843Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"doi_compliance","ran_at":"2026-05-19T12:33:31.546986Z","status":"completed","version":"1.0.0","findings_count":0}],"snapshot_sha256":"162ddda1bbe6f2db18cebf495266973fcf3756ba02a57ed1f4fd0e4d62967ed0"},"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"},"verdict_id":"c6a1c7da-9796-4d61-893c-c22d84397521"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-22T02:04:41Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"ltfWL0H8H4QYQMMKOJzykuXqAqk/7dKAHHDzmUeUY/sGxKn/8g7BNEmw37Y8OHmaNGCRmKp+/WL03xoXGLeJCQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-23T12:10:40.740467Z"},"content_sha256":"0c1c81e7c89acdacd33cf45a98c309b015c5caa18c6cdb2d945468d3de6ecdd9","schema_version":"1.0","event_id":"sha256:0c1c81e7c89acdacd33cf45a98c309b015c5caa18c6cdb2d945468d3de6ecdd9"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/7XEA25FQG2QRI2FNXXF7MDASVD/bundle.json","state_url":"https://pith.science/pith/7XEA25FQG2QRI2FNXXF7MDASVD/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/7XEA25FQG2QRI2FNXXF7MDASVD/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":"2026-05-23T12:10:40Z","links":{"resolver":"https://pith.science/pith/7XEA25FQG2QRI2FNXXF7MDASVD","bundle":"https://pith.science/pith/7XEA25FQG2QRI2FNXXF7MDASVD/bundle.json","state":"https://pith.science/pith/7XEA25FQG2QRI2FNXXF7MDASVD/state.json","well_known_bundle":"https://pith.science/.well-known/pith/7XEA25FQG2QRI2FNXXF7MDASVD/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:7XEA25FQG2QRI2FNXXF7MDASVD","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":"609fb5e8e5266b7b8008b4f1e30100c9dc1b5b20f5eb31361abcd749c2054462","cross_cats_sorted":["cs.AI","cs.LG"],"license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.CL","submitted_at":"2026-05-07T17:22:11Z","title_canon_sha256":"9ee1af973d04efab55c7bc8c341184da8f541824251ef46afd93b72e02342787"},"schema_version":"1.0","source":{"id":"2605.06597","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.06597","created_at":"2026-05-22T02:04:41Z"},{"alias_kind":"arxiv_version","alias_value":"2605.06597v2","created_at":"2026-05-22T02:04:41Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.06597","created_at":"2026-05-22T02:04:41Z"},{"alias_kind":"pith_short_12","alias_value":"7XEA25FQG2QR","created_at":"2026-05-22T02:04:41Z"},{"alias_kind":"pith_short_16","alias_value":"7XEA25FQG2QRI2FN","created_at":"2026-05-22T02:04:41Z"},{"alias_kind":"pith_short_8","alias_value":"7XEA25FQ","created_at":"2026-05-22T02:04:41Z"}],"graph_snapshots":[{"event_id":"sha256:0c1c81e7c89acdacd33cf45a98c309b015c5caa18c6cdb2d945468d3de6ecdd9","target":"graph","created_at":"2026-05-22T02:04:41Z","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":"UniSDfull, an integrated pipeline that combines complementary components, achieves the strongest overall performance, improving over the base model by +5.4 points and the strongest baseline by +2.8 points."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"That the listed mechanisms (multi-teacher agreement, EMA, token contrastive learning, feature matching, divergence clipping) reliably address supervision instability in free-form self-generated trajectories and that their interactions can be isolated and combined without post-hoc selection bias affecting the reported gains."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"UniSD unifies complementary self-distillation mechanisms for autoregressive LLMs and achieves up to +5.4 point gains over base models and +2.8 over baselines across six benchmarks and six models."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"A unified framework makes self-distillation a reliable way to adapt large language models without stronger teachers."}],"snapshot_sha256":"43f0993ffab088ed0c3a24d3818d0127f6c5b5d58b2ff8847fd9c854ec42b87a"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"integrity":{"available":true,"clean":true,"detectors_run":[{"findings_count":0,"name":"claim_evidence","ran_at":"2026-05-20T12:22:03.785641Z","status":"completed","version":"1.0.0"},{"findings_count":0,"name":"ai_meta_artifact","ran_at":"2026-05-20T07:37:52.552364Z","status":"completed","version":"1.0.0"},{"findings_count":0,"name":"doi_title_agreement","ran_at":"2026-05-19T18:01:19.364843Z","status":"completed","version":"1.0.0"},{"findings_count":0,"name":"doi_compliance","ran_at":"2026-05-19T12:33:31.546986Z","status":"completed","version":"1.0.0"}],"endpoint":"/pith/2605.06597/integrity.json","findings":[],"snapshot_sha256":"162ddda1bbe6f2db18cebf495266973fcf3756ba02a57ed1f4fd0e4d62967ed0","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Self-distillation (SD) offers a promising path for adapting large language models (LLMs) without relying on stronger external teachers. However, SD in autoregressive LLMs remains challenging because self-generated trajectories are free-form, correctness is task-dependent, and plausible rationales can still provide unstable or unreliable supervision. Existing methods mainly examine isolated design choices, leaving their effectiveness, roles, and interactions unclear. In this paper, we propose UniSD, a unified framework to systematically study self-distillation. UniSD integrates complementary me","authors_text":"B. Aditya Prakash, Haoxin Liu, Jindong Wang, Josiah Hester, Lucheng Fu, Srijan Kumar, Yijia Xiao, Yinyi Luo, Yiqiao Jin, Yiyang Wang","cross_cats":["cs.AI","cs.LG"],"headline":"A unified framework makes self-distillation a reliable way to adapt large language models without stronger teachers.","license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.CL","submitted_at":"2026-05-07T17:22:11Z","title":"UniSD: Towards a Unified Self-Distillation Framework for Large Language Models"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.06597","kind":"arxiv","version":2},"verdict":{"created_at":"2026-05-08T09:52:45.380637Z","id":"c6a1c7da-9796-4d61-893c-c22d84397521","model_set":{"reader":"grok-4.3"},"one_line_summary":"UniSD unifies complementary self-distillation mechanisms for autoregressive LLMs and achieves up to +5.4 point gains over base models and +2.8 over baselines across six benchmarks and six models.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"A unified framework makes self-distillation a reliable way to adapt large language models without stronger teachers.","strongest_claim":"UniSDfull, an integrated pipeline that combines complementary components, achieves the strongest overall performance, improving over the base model by +5.4 points and the strongest baseline by +2.8 points.","weakest_assumption":"That the listed mechanisms (multi-teacher agreement, EMA, token contrastive learning, feature matching, divergence clipping) reliably address supervision instability in free-form self-generated trajectories and that their interactions can be isolated and combined without post-hoc selection bias affecting the reported gains."}},"verdict_id":"c6a1c7da-9796-4d61-893c-c22d84397521"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:c659f9c633dfbdf3d883114d5456ce7d2121bc5bda13650907229d2cdfebf559","target":"record","created_at":"2026-05-22T02:04:41Z","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":"609fb5e8e5266b7b8008b4f1e30100c9dc1b5b20f5eb31361abcd749c2054462","cross_cats_sorted":["cs.AI","cs.LG"],"license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.CL","submitted_at":"2026-05-07T17:22:11Z","title_canon_sha256":"9ee1af973d04efab55c7bc8c341184da8f541824251ef46afd93b72e02342787"},"schema_version":"1.0","source":{"id":"2605.06597","kind":"arxiv","version":2}},"canonical_sha256":"fdc80d74b036a11468adbdcbf60c12a8fba5c294aa7459c60fd9f78cb640d93d","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"fdc80d74b036a11468adbdcbf60c12a8fba5c294aa7459c60fd9f78cb640d93d","first_computed_at":"2026-05-22T02:04:41.944621Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-22T02:04:41.944621Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"gyF1yaXC3sPgmGSU6uyD6hY02kKytmu0mpyGTpZF+34YCgIDm6RgqMPnvAyCYwwIitOM/NliQncLk8k2SjIBCQ==","signature_status":"signed_v1","signed_at":"2026-05-22T02:04:41.945423Z","signed_message":"canonical_sha256_bytes"},"source_id":"2605.06597","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:c659f9c633dfbdf3d883114d5456ce7d2121bc5bda13650907229d2cdfebf559","sha256:0c1c81e7c89acdacd33cf45a98c309b015c5caa18c6cdb2d945468d3de6ecdd9"],"state_sha256":"ab8cac07d642ffee95228bdad0942799a5ec5df289e5a9f4505099ba3435481d"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"NL+BB7N9Yd2R/ILCElOf3iK6KCKDWnVE7Egposv1YiQPmL1Nt5YR9QTIiOdu+PFmoS44ZHg1r0Hfnw9yD13BCg==","signed_message":"bundle_sha256_bytes","signed_at":"2026-05-23T12:10:40.745011Z","bundle_sha256":"1aa91f4545370603c9506edae777c0b8c495b774973e83b1d2ec7c809297f9a8"}}