{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:JBDOFT6L6MFKATS7ER4ZR5DAEI","short_pith_number":"pith:JBDOFT6L","schema_version":"1.0","canonical_sha256":"4846e2cfcbf30aa04e5f247998f460222be686b23dcee0b22abe10779852e2ba","source":{"kind":"arxiv","id":"2605.25244","version":1},"attestation_state":"computed","paper":{"title":"Inference Time Optimization with Confidence Dynamics","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Ankit Shah, Jiayun Wang, Jinrui Huang, Minghao Liu, Wei Wei, Yu Wang","submitted_at":"2026-05-24T20:04:19Z","abstract_excerpt":"Inference time optimization techniques, such as repeated sampling, have significantly advanced the reasoning capabilities of Large Language Models (LLMs). However, the critical role of model uncertainty remains largely underexplored in these optimization strategies. In this paper, we investigate the dynamics of confidence along reasoning trajectories and for first time reveal a surprising and unique pattern: correct answer traces tend to exhibit confidence improvement over time (positive confidence gain), while incorrect traces show attenuated or declining confidence as reasoning proceeds. Bas"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2605.25244","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2026-05-24T20:04:19Z","cross_cats_sorted":[],"title_canon_sha256":"56ac2c9ef2e11ab47fb3ed791111538a6f95287154bfc7f10cc062e0088cc549","abstract_canon_sha256":"6169bd924b97321622cda3b8b7cb0107c195399c55ebc9c00e4846e8f104e1ed"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-26T02:04:25.031942Z","signature_b64":"WbgRCD64xf2LDN1h7XjTglinI3/Vp3FQ3sgJKK1opYbnM+Ak1/jTn3e11SZno72DgfsPYmYaL1wD/AIHlY5wBw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"4846e2cfcbf30aa04e5f247998f460222be686b23dcee0b22abe10779852e2ba","last_reissued_at":"2026-05-26T02:04:25.031103Z","signature_status":"signed_v1","first_computed_at":"2026-05-26T02:04:25.031103Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Inference Time Optimization with Confidence Dynamics","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Ankit Shah, Jiayun Wang, Jinrui Huang, Minghao Liu, Wei Wei, Yu Wang","submitted_at":"2026-05-24T20:04:19Z","abstract_excerpt":"Inference time optimization techniques, such as repeated sampling, have significantly advanced the reasoning capabilities of Large Language Models (LLMs). However, the critical role of model uncertainty remains largely underexplored in these optimization strategies. In this paper, we investigate the dynamics of confidence along reasoning trajectories and for first time reveal a surprising and unique pattern: correct answer traces tend to exhibit confidence improvement over time (positive confidence gain), while incorrect traces show attenuated or declining confidence as reasoning proceeds. Bas"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.25244","kind":"arxiv","version":1},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2605.25244/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"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"},"aliases":[{"alias_kind":"arxiv","alias_value":"2605.25244","created_at":"2026-05-26T02:04:25.031255+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.25244v1","created_at":"2026-05-26T02:04:25.031255+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.25244","created_at":"2026-05-26T02:04:25.031255+00:00"},{"alias_kind":"pith_short_12","alias_value":"JBDOFT6L6MFK","created_at":"2026-05-26T02:04:25.031255+00:00"},{"alias_kind":"pith_short_16","alias_value":"JBDOFT6L6MFKATS7","created_at":"2026-05-26T02:04:25.031255+00:00"},{"alias_kind":"pith_short_8","alias_value":"JBDOFT6L","created_at":"2026-05-26T02:04:25.031255+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/JBDOFT6L6MFKATS7ER4ZR5DAEI","json":"https://pith.science/pith/JBDOFT6L6MFKATS7ER4ZR5DAEI.json","graph_json":"https://pith.science/api/pith-number/JBDOFT6L6MFKATS7ER4ZR5DAEI/graph.json","events_json":"https://pith.science/api/pith-number/JBDOFT6L6MFKATS7ER4ZR5DAEI/events.json","paper":"https://pith.science/paper/JBDOFT6L"},"agent_actions":{"view_html":"https://pith.science/pith/JBDOFT6L6MFKATS7ER4ZR5DAEI","download_json":"https://pith.science/pith/JBDOFT6L6MFKATS7ER4ZR5DAEI.json","view_paper":"https://pith.science/paper/JBDOFT6L","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.25244&json=true","fetch_graph":"https://pith.science/api/pith-number/JBDOFT6L6MFKATS7ER4ZR5DAEI/graph.json","fetch_events":"https://pith.science/api/pith-number/JBDOFT6L6MFKATS7ER4ZR5DAEI/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/JBDOFT6L6MFKATS7ER4ZR5DAEI/action/timestamp_anchor","attest_storage":"https://pith.science/pith/JBDOFT6L6MFKATS7ER4ZR5DAEI/action/storage_attestation","attest_author":"https://pith.science/pith/JBDOFT6L6MFKATS7ER4ZR5DAEI/action/author_attestation","sign_citation":"https://pith.science/pith/JBDOFT6L6MFKATS7ER4ZR5DAEI/action/citation_signature","submit_replication":"https://pith.science/pith/JBDOFT6L6MFKATS7ER4ZR5DAEI/action/replication_record"}},"created_at":"2026-05-26T02:04:25.031255+00:00","updated_at":"2026-05-26T02:04:25.031255+00:00"}