{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2026:EWJ3PILCTYJ5P75WZHXBIPY6L2","short_pith_number":"pith:EWJ3PILC","canonical_record":{"source":{"id":"2605.14457","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2026-05-14T06:52:59Z","cross_cats_sorted":[],"title_canon_sha256":"e9ae494600d475e69e783e533bca6dbe654e76650bf7186efc7b220f4352fd04","abstract_canon_sha256":"bacb333a818aac7abf9550ff7cd7519b4c32100efb9fae4cc3fed10a0d728184"},"schema_version":"1.0"},"canonical_sha256":"2593b7a1629e13d7ffb6c9ee143f1e5eba4e57083221e96f0c297a1eea4df5a9","source":{"kind":"arxiv","id":"2605.14457","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.14457","created_at":"2026-05-17T23:39:06Z"},{"alias_kind":"arxiv_version","alias_value":"2605.14457v1","created_at":"2026-05-17T23:39:06Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.14457","created_at":"2026-05-17T23:39:06Z"},{"alias_kind":"pith_short_12","alias_value":"EWJ3PILCTYJ5","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_16","alias_value":"EWJ3PILCTYJ5P75W","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_8","alias_value":"EWJ3PILC","created_at":"2026-05-18T12:33:37Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2026:EWJ3PILCTYJ5P75WZHXBIPY6L2","target":"record","payload":{"canonical_record":{"source":{"id":"2605.14457","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2026-05-14T06:52:59Z","cross_cats_sorted":[],"title_canon_sha256":"e9ae494600d475e69e783e533bca6dbe654e76650bf7186efc7b220f4352fd04","abstract_canon_sha256":"bacb333a818aac7abf9550ff7cd7519b4c32100efb9fae4cc3fed10a0d728184"},"schema_version":"1.0"},"canonical_sha256":"2593b7a1629e13d7ffb6c9ee143f1e5eba4e57083221e96f0c297a1eea4df5a9","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-17T23:39:06.826059Z","signature_b64":"CYfOA+9RDW8j5yIHDfdERLHbvhI+RuhOvlhiM7LgG2IEWDcxA+EFZhtikl5C4VTk+L1G45JbvP9M4sE15B4+AA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"2593b7a1629e13d7ffb6c9ee143f1e5eba4e57083221e96f0c297a1eea4df5a9","last_reissued_at":"2026-05-17T23:39:06.825383Z","signature_status":"signed_v1","first_computed_at":"2026-05-17T23:39:06.825383Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2605.14457","source_version":1,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-17T23:39:06Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"9auqL6pbvbhda7zwL/7btiTgWo00VFLAFUKVUqBJh8qjeemdr1G6IHbI2T1oQLXbX5ieJ6UMDDQLFnifbSr8AA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-06-11T04:59:37.242632Z"},"content_sha256":"0fa3d2b55afb33731e29fe8f0d5ad17389114d0fdd3214225089d2dd20809f1f","schema_version":"1.0","event_id":"sha256:0fa3d2b55afb33731e29fe8f0d5ad17389114d0fdd3214225089d2dd20809f1f"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2026:EWJ3PILCTYJ5P75WZHXBIPY6L2","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Stateful Reasoning via Insight Replay","license":"http://creativecommons.org/licenses/by/4.0/","headline":"Replaying critical insights from earlier in a reasoning trace keeps them accessible and improves accuracy as chains lengthen in large language models.","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Ang Li, Bin Lei, Caiwen Ding, Jiachen Yang, Xin Eric Wang","submitted_at":"2026-05-14T06:52:59Z","abstract_excerpt":"Chain-of-Thought (CoT) reasoning has become a foundation for eliciting multi-step reasoning in large language models, but recent studies show that its benefits do not scale monotonically with chain length: while longer CoT generally enables a model to tackle harder problems, on a given problem, accuracy typically increases with CoT length up to a point, after which it declines. We identify a major cause of this phenomenon: as the CoT grows, the model's attention to critical insights produced earlier in the trace gradually weakens, making those insights progressively less accessible when they a"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"3-round InsightReplay yields accuracy gains across all 24 settings, with an averaged improvement of +1.65 points over standard CoT, and a largest single-setting gain of +9.2 points on R1-Distill-32B's LiveCodeBench v5 subset.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That the model can reliably extract truly critical insights without introducing noise or errors, and that replaying them will restore accessibility without disrupting the ongoing reasoning flow.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"InsightReplay improves LLM accuracy on reasoning benchmarks by extracting and replaying critical insights to maintain their accessibility during extended chain-of-thought generation.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Replaying critical insights from earlier in a reasoning trace keeps them accessible and improves accuracy as chains lengthen in large language models.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"5d616a79046304b9c8d7c649f194be86cd1ec7d3ab6967f11dd5a681a3d99935"},"source":{"id":"2605.14457","kind":"arxiv","version":1},"verdict":{"id":"a18e5c72-95d6-4958-95ba-58d245d5811d","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-15T01:37:03.082135Z","strongest_claim":"3-round InsightReplay yields accuracy gains across all 24 settings, with an averaged improvement of +1.65 points over standard CoT, and a largest single-setting gain of +9.2 points on R1-Distill-32B's LiveCodeBench v5 subset.","one_line_summary":"InsightReplay improves LLM accuracy on reasoning benchmarks by extracting and replaying critical insights to maintain their accessibility during extended chain-of-thought generation.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That the model can reliably extract truly critical insights without introducing noise or errors, and that replaying them will restore accessibility without disrupting the ongoing reasoning flow.","pith_extraction_headline":"Replaying critical insights from earlier in a reasoning trace keeps them accessible and improves accuracy as chains lengthen in large language models."},"references":{"count":37,"sample":[{"doi":"","year":2022,"title":"Chain-of-thought prompting elicits reasoning in large language models.Advances in neural information processing systems, 35:24824–24837","work_id":"3256b5a6-76d9-460c-ad77-5d232058ad6d","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2023,"title":"Self-consistency improves chain of thought reasoning in language models","work_id":"bc8f4e8c-9c0a-4b85-8532-6489ef6b2d1c","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2025,"title":"Deepseek-r1 incentivizes reasoning in llms through reinforcement learning.Nature, 645(8081):633–638","work_id":"f01253de-58e9-4b0f-a91e-4ef797517a24","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2022,"title":"Large language models are zero-shot reasoners.Advances in neural information processing systems, 35: 22199–22213","work_id":"f69f3f9c-f092-49b0-b244-16b6480f5a02","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2025,"title":"When more is less: Understanding chain-of-thought length in llms","work_id":"450a518d-6378-4944-9651-8e3ef8049ddf","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":37,"snapshot_sha256":"7fe9431531b7d685697d2fb2b44b6a06ea83eafa258a7cdc5f8e4d20a81f320b","internal_anchors":6},"formal_canon":{"evidence_count":2,"snapshot_sha256":"ffc630b03d1266ec44cbb25299b9d66d1fd6dc53e47df7df2f2e81917d3e240e"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"verdict_id":"a18e5c72-95d6-4958-95ba-58d245d5811d"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-17T23:39:06Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"tIAkBhIahCi97Dd5R1a7+D7ltcP2/98r8llUumrzIQuV27R5k/G6KzE5JLuaB8wwZ7vBi9ToD5g8HB037Cb0DQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-06-11T04:59:37.243870Z"},"content_sha256":"7509ac2ed158ccf031aaa5858a2ec6df7303256ec0d383e0a962f26708129dc1","schema_version":"1.0","event_id":"sha256:7509ac2ed158ccf031aaa5858a2ec6df7303256ec0d383e0a962f26708129dc1"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/EWJ3PILCTYJ5P75WZHXBIPY6L2/bundle.json","state_url":"https://pith.science/pith/EWJ3PILCTYJ5P75WZHXBIPY6L2/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/EWJ3PILCTYJ5P75WZHXBIPY6L2/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":"2026-06-11T04:59:37Z","links":{"resolver":"https://pith.science/pith/EWJ3PILCTYJ5P75WZHXBIPY6L2","bundle":"https://pith.science/pith/EWJ3PILCTYJ5P75WZHXBIPY6L2/bundle.json","state":"https://pith.science/pith/EWJ3PILCTYJ5P75WZHXBIPY6L2/state.json","well_known_bundle":"https://pith.science/.well-known/pith/EWJ3PILCTYJ5P75WZHXBIPY6L2/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:EWJ3PILCTYJ5P75WZHXBIPY6L2","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":"bacb333a818aac7abf9550ff7cd7519b4c32100efb9fae4cc3fed10a0d728184","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2026-05-14T06:52:59Z","title_canon_sha256":"e9ae494600d475e69e783e533bca6dbe654e76650bf7186efc7b220f4352fd04"},"schema_version":"1.0","source":{"id":"2605.14457","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.14457","created_at":"2026-05-17T23:39:06Z"},{"alias_kind":"arxiv_version","alias_value":"2605.14457v1","created_at":"2026-05-17T23:39:06Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.14457","created_at":"2026-05-17T23:39:06Z"},{"alias_kind":"pith_short_12","alias_value":"EWJ3PILCTYJ5","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_16","alias_value":"EWJ3PILCTYJ5P75W","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_8","alias_value":"EWJ3PILC","created_at":"2026-05-18T12:33:37Z"}],"graph_snapshots":[{"event_id":"sha256:7509ac2ed158ccf031aaa5858a2ec6df7303256ec0d383e0a962f26708129dc1","target":"graph","created_at":"2026-05-17T23:39:06Z","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":"3-round InsightReplay yields accuracy gains across all 24 settings, with an averaged improvement of +1.65 points over standard CoT, and a largest single-setting gain of +9.2 points on R1-Distill-32B's LiveCodeBench v5 subset."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"That the model can reliably extract truly critical insights without introducing noise or errors, and that replaying them will restore accessibility without disrupting the ongoing reasoning flow."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"InsightReplay improves LLM accuracy on reasoning benchmarks by extracting and replaying critical insights to maintain their accessibility during extended chain-of-thought generation."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"Replaying critical insights from earlier in a reasoning trace keeps them accessible and improves accuracy as chains lengthen in large language models."}],"snapshot_sha256":"5d616a79046304b9c8d7c649f194be86cd1ec7d3ab6967f11dd5a681a3d99935"},"formal_canon":{"evidence_count":2,"snapshot_sha256":"ffc630b03d1266ec44cbb25299b9d66d1fd6dc53e47df7df2f2e81917d3e240e"},"paper":{"abstract_excerpt":"Chain-of-Thought (CoT) reasoning has become a foundation for eliciting multi-step reasoning in large language models, but recent studies show that its benefits do not scale monotonically with chain length: while longer CoT generally enables a model to tackle harder problems, on a given problem, accuracy typically increases with CoT length up to a point, after which it declines. We identify a major cause of this phenomenon: as the CoT grows, the model's attention to critical insights produced earlier in the trace gradually weakens, making those insights progressively less accessible when they a","authors_text":"Ang Li, Bin Lei, Caiwen Ding, Jiachen Yang, Xin Eric Wang","cross_cats":[],"headline":"Replaying critical insights from earlier in a reasoning trace keeps them accessible and improves accuracy as chains lengthen in large language models.","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2026-05-14T06:52:59Z","title":"Stateful Reasoning via Insight Replay"},"references":{"count":37,"internal_anchors":6,"resolved_work":37,"sample":[{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":1,"title":"Chain-of-thought prompting elicits reasoning in large language models.Advances in neural information processing systems, 35:24824–24837","work_id":"3256b5a6-76d9-460c-ad77-5d232058ad6d","year":2022},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":2,"title":"Self-consistency improves chain of thought reasoning in language models","work_id":"bc8f4e8c-9c0a-4b85-8532-6489ef6b2d1c","year":2023},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":3,"title":"Deepseek-r1 incentivizes reasoning in llms through reinforcement learning.Nature, 645(8081):633–638","work_id":"f01253de-58e9-4b0f-a91e-4ef797517a24","year":2025},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":4,"title":"Large language models are zero-shot reasoners.Advances in neural information processing systems, 35: 22199–22213","work_id":"f69f3f9c-f092-49b0-b244-16b6480f5a02","year":2022},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":5,"title":"When more is less: Understanding chain-of-thought length in llms","work_id":"450a518d-6378-4944-9651-8e3ef8049ddf","year":2025}],"snapshot_sha256":"7fe9431531b7d685697d2fb2b44b6a06ea83eafa258a7cdc5f8e4d20a81f320b"},"source":{"id":"2605.14457","kind":"arxiv","version":1},"verdict":{"created_at":"2026-05-15T01:37:03.082135Z","id":"a18e5c72-95d6-4958-95ba-58d245d5811d","model_set":{"reader":"grok-4.3"},"one_line_summary":"InsightReplay improves LLM accuracy on reasoning benchmarks by extracting and replaying critical insights to maintain their accessibility during extended chain-of-thought generation.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"Replaying critical insights from earlier in a reasoning trace keeps them accessible and improves accuracy as chains lengthen in large language models.","strongest_claim":"3-round InsightReplay yields accuracy gains across all 24 settings, with an averaged improvement of +1.65 points over standard CoT, and a largest single-setting gain of +9.2 points on R1-Distill-32B's LiveCodeBench v5 subset.","weakest_assumption":"That the model can reliably extract truly critical insights without introducing noise or errors, and that replaying them will restore accessibility without disrupting the ongoing reasoning flow."}},"verdict_id":"a18e5c72-95d6-4958-95ba-58d245d5811d"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:0fa3d2b55afb33731e29fe8f0d5ad17389114d0fdd3214225089d2dd20809f1f","target":"record","created_at":"2026-05-17T23:39:06Z","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":"bacb333a818aac7abf9550ff7cd7519b4c32100efb9fae4cc3fed10a0d728184","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2026-05-14T06:52:59Z","title_canon_sha256":"e9ae494600d475e69e783e533bca6dbe654e76650bf7186efc7b220f4352fd04"},"schema_version":"1.0","source":{"id":"2605.14457","kind":"arxiv","version":1}},"canonical_sha256":"2593b7a1629e13d7ffb6c9ee143f1e5eba4e57083221e96f0c297a1eea4df5a9","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"2593b7a1629e13d7ffb6c9ee143f1e5eba4e57083221e96f0c297a1eea4df5a9","first_computed_at":"2026-05-17T23:39:06.825383Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-17T23:39:06.825383Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"CYfOA+9RDW8j5yIHDfdERLHbvhI+RuhOvlhiM7LgG2IEWDcxA+EFZhtikl5C4VTk+L1G45JbvP9M4sE15B4+AA==","signature_status":"signed_v1","signed_at":"2026-05-17T23:39:06.826059Z","signed_message":"canonical_sha256_bytes"},"source_id":"2605.14457","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:0fa3d2b55afb33731e29fe8f0d5ad17389114d0fdd3214225089d2dd20809f1f","sha256:7509ac2ed158ccf031aaa5858a2ec6df7303256ec0d383e0a962f26708129dc1"],"state_sha256":"bdc2c6f7b2beaf47642119e76cc838ab53d5674be444e9284b88fab06d395377"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"9voiaIJLYY0ste8sHwirpa+vCO2pi45CbZXKAaN0QrlZGyUZg54SU4dwPD9zFJlHAEgPsJtjh6s7IcxBOFEoBQ==","signed_message":"bundle_sha256_bytes","signed_at":"2026-06-11T04:59:37.249446Z","bundle_sha256":"ecf77ac2be352040ba26ce276c851f86533c9a7b1d42330b312ce3ab101ffff6"}}