{"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"}