{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:5CB62F5S426VGG2FZUHOBOMXII","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":"dece95c2b7e1c506da16d2c512b9e0bf6c3b4fd1ef430486a5d981067387d028","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-08-21T05:48:38Z","title_canon_sha256":"d51da53a66a1207a671bc95f9f5b03e63ebb297d7d46f939aea43897eac3ed8c"},"schema_version":"1.0","source":{"id":"2508.15260","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2508.15260","created_at":"2026-05-17T23:38:48Z"},{"alias_kind":"arxiv_version","alias_value":"2508.15260v1","created_at":"2026-05-17T23:38:48Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2508.15260","created_at":"2026-05-17T23:38:48Z"},{"alias_kind":"pith_short_12","alias_value":"5CB62F5S426V","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_16","alias_value":"5CB62F5S426VGG2F","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_8","alias_value":"5CB62F5S","created_at":"2026-05-18T12:33:37Z"}],"graph_snapshots":[{"event_id":"sha256:d7796c2b1deef7df910f8353f9ba468164b40e555a2c1b7b7d4bb207889b933c","target":"graph","created_at":"2026-05-17T23:38:48Z","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":"on challenging benchmarks such as AIME 2025, DeepConf@512 achieves up to 99.9% accuracy and reduces generated tokens by up to 84.7% compared to full parallel thinking."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"That model-internal confidence signals reliably indicate the quality or correctness of individual reasoning traces, allowing effective dynamic filtering without additional training or tuning."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"DeepConf filters low-quality reasoning traces using model-internal confidence signals, reaching up to 99.9% accuracy on AIME 2025 while reducing generated tokens by up to 84.7% versus full parallel sampling."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"DeepConf uses model-internal confidence to filter weak reasoning traces, reaching 99.9% accuracy on AIME 2025 while cutting tokens by 84.7%."}],"snapshot_sha256":"5db56bd422fe8c58ed9b577bcb5bc786e62c63c983f183ad154af04a9e6a406d"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"paper":{"abstract_excerpt":"Large Language Models (LLMs) have shown great potential in reasoning tasks through test-time scaling methods like self-consistency with majority voting. However, this approach often leads to diminishing returns in accuracy and high computational overhead. To address these challenges, we introduce Deep Think with Confidence (DeepConf), a simple yet powerful method that enhances both reasoning efficiency and performance at test time. DeepConf leverages model-internal confidence signals to dynamically filter out low-quality reasoning traces during or after generation. It requires no additional mo","authors_text":"Jiawei Zhao, Xuewei Wang, Yichao Fu, Yuandong Tian","cross_cats":[],"headline":"DeepConf uses model-internal confidence to filter weak reasoning traces, reaching 99.9% accuracy on AIME 2025 while cutting tokens by 84.7%.","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-08-21T05:48:38Z","title":"Deep Think with Confidence"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2508.15260","kind":"arxiv","version":1},"verdict":{"created_at":"2026-05-16T11:26:17.715523Z","id":"7485b2b8-cca9-4e0f-ab75-e669a96c37ed","model_set":{"reader":"grok-4.3"},"one_line_summary":"DeepConf filters low-quality reasoning traces using model-internal confidence signals, reaching up to 99.9% accuracy on AIME 2025 while reducing generated tokens by up to 84.7% versus full parallel sampling.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"DeepConf uses model-internal confidence to filter weak reasoning traces, reaching 99.9% accuracy on AIME 2025 while cutting tokens by 84.7%.","strongest_claim":"on challenging benchmarks such as AIME 2025, DeepConf@512 achieves up to 99.9% accuracy and reduces generated tokens by up to 84.7% compared to full parallel thinking.","weakest_assumption":"That model-internal confidence signals reliably indicate the quality or correctness of individual reasoning traces, allowing effective dynamic filtering without additional training or tuning."}},"verdict_id":"7485b2b8-cca9-4e0f-ab75-e669a96c37ed"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:e3c287cccf50cd6d48c5e8d5d74853e4e5616793e4fb77a36f2df21ed5a4d869","target":"record","created_at":"2026-05-17T23:38:48Z","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":"dece95c2b7e1c506da16d2c512b9e0bf6c3b4fd1ef430486a5d981067387d028","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-08-21T05:48:38Z","title_canon_sha256":"d51da53a66a1207a671bc95f9f5b03e63ebb297d7d46f939aea43897eac3ed8c"},"schema_version":"1.0","source":{"id":"2508.15260","kind":"arxiv","version":1}},"canonical_sha256":"e883ed17b2e6bd531b45cd0ee0b9974214301e7121f8e8ad21dab941de352c22","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"e883ed17b2e6bd531b45cd0ee0b9974214301e7121f8e8ad21dab941de352c22","first_computed_at":"2026-05-17T23:38:48.026210Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-17T23:38:48.026210Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"1U1+34h5nugLI0uJRSixwlRnquhLcD1K26XjxdvmSyncamd3xjVWHbArvPSlUzy6XS0n/kfR69qglerqUJMTAQ==","signature_status":"signed_v1","signed_at":"2026-05-17T23:38:48.026780Z","signed_message":"canonical_sha256_bytes"},"source_id":"2508.15260","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:e3c287cccf50cd6d48c5e8d5d74853e4e5616793e4fb77a36f2df21ed5a4d869","sha256:d7796c2b1deef7df910f8353f9ba468164b40e555a2c1b7b7d4bb207889b933c"],"state_sha256":"7918ea6b7e62cca19e6c160c7caf1827dae733dc9803c1929d4c1be5472c835d"}