{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:KB5AWN5QHUBGG43C6MF5BSVUEO","short_pith_number":"pith:KB5AWN5Q","schema_version":"1.0","canonical_sha256":"507a0b37b03d02637362f30bd0cab423a899f8dde7e80f495ae7f768d0b61a78","source":{"kind":"arxiv","id":"2608.00605","version":1},"attestation_state":"computed","paper":{"title":"Escaping Confidence Trap: Evolutionary Decoding for Mathematical Reasoning in Diffusion LLMs","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Dacheng Tao, Daoyi Dong, Davide Modolo, Hanqing Zhao, Jue Wang, Liuyue Xie, Rongcheng Tu, Xu Zhang, Yatao Bian, Zhenhong Sun","submitted_at":"2026-08-01T11:48:25Z","abstract_excerpt":"Diffusion large language models (dLLMs) have emerged as a promising alternative to autoregressive LLMs, offering efficient generation through block-wise progressive unmasking. However, their strong general-purpose performance does not necessarily translate into reliable mathematical reasoning, where correctness depends on preserving coherent numerical-symbolic reasoning trajectories. In this work, we analyze the decoding trajectories of LLaDA 2.0 and identify a recurring diffusion confidence trap: local token confidence can become misaligned with global reasoning correctness during progressive"},"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":"2608.00605","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.AI","submitted_at":"2026-08-01T11:48:25Z","cross_cats_sorted":[],"title_canon_sha256":"42fe755fb3599d6d5c8df6171937461aa8b375b4cf1a3699ba1ca43051d87942","abstract_canon_sha256":"5dbc6346812f7183c86cf2b01dfd65ccc66c434e0fd4f0a3907f9632612454ad"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-08-04T00:38:04.288865Z","signature_b64":"Pd8MdIca0Vkw7sSJnV5TX+v8nmAJgUZwLh+SYJZEQKlUI/d8fZ3ZPqxjaoxUp5Ke/QZ1WZS17Cl0ufdciZj5Ag==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"507a0b37b03d02637362f30bd0cab423a899f8dde7e80f495ae7f768d0b61a78","last_reissued_at":"2026-08-04T00:38:04.287427Z","signature_status":"signed_v1","first_computed_at":"2026-08-04T00:38:04.287427Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Escaping Confidence Trap: Evolutionary Decoding for Mathematical Reasoning in Diffusion LLMs","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Dacheng Tao, Daoyi Dong, Davide Modolo, Hanqing Zhao, Jue Wang, Liuyue Xie, Rongcheng Tu, Xu Zhang, Yatao Bian, Zhenhong Sun","submitted_at":"2026-08-01T11:48:25Z","abstract_excerpt":"Diffusion large language models (dLLMs) have emerged as a promising alternative to autoregressive LLMs, offering efficient generation through block-wise progressive unmasking. However, their strong general-purpose performance does not necessarily translate into reliable mathematical reasoning, where correctness depends on preserving coherent numerical-symbolic reasoning trajectories. In this work, we analyze the decoding trajectories of LLaDA 2.0 and identify a recurring diffusion confidence trap: local token confidence can become misaligned with global reasoning correctness during progressive"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2608.00605","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/2608.00605/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":"2608.00605","created_at":"2026-08-04T00:38:04.288626+00:00"},{"alias_kind":"arxiv_version","alias_value":"2608.00605v1","created_at":"2026-08-04T00:38:04.288626+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2608.00605","created_at":"2026-08-04T00:38:04.288626+00:00"},{"alias_kind":"pith_short_12","alias_value":"KB5AWN5QHUBG","created_at":"2026-08-04T00:38:04.288626+00:00"},{"alias_kind":"pith_short_16","alias_value":"KB5AWN5QHUBGG43C","created_at":"2026-08-04T00:38:04.288626+00:00"},{"alias_kind":"pith_short_8","alias_value":"KB5AWN5Q","created_at":"2026-08-04T00:38:04.288626+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/KB5AWN5QHUBGG43C6MF5BSVUEO","json":"https://pith.science/pith/KB5AWN5QHUBGG43C6MF5BSVUEO.json","graph_json":"https://pith.science/api/pith-number/KB5AWN5QHUBGG43C6MF5BSVUEO/graph.json","events_json":"https://pith.science/api/pith-number/KB5AWN5QHUBGG43C6MF5BSVUEO/events.json","paper":"https://pith.science/paper/KB5AWN5Q"},"agent_actions":{"view_html":"https://pith.science/pith/KB5AWN5QHUBGG43C6MF5BSVUEO","download_json":"https://pith.science/pith/KB5AWN5QHUBGG43C6MF5BSVUEO.json","view_paper":"https://pith.science/paper/KB5AWN5Q","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2608.00605&json=true","fetch_graph":"https://pith.science/api/pith-number/KB5AWN5QHUBGG43C6MF5BSVUEO/graph.json","fetch_events":"https://pith.science/api/pith-number/KB5AWN5QHUBGG43C6MF5BSVUEO/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/KB5AWN5QHUBGG43C6MF5BSVUEO/action/timestamp_anchor","attest_storage":"https://pith.science/pith/KB5AWN5QHUBGG43C6MF5BSVUEO/action/storage_attestation","attest_author":"https://pith.science/pith/KB5AWN5QHUBGG43C6MF5BSVUEO/action/author_attestation","sign_citation":"https://pith.science/pith/KB5AWN5QHUBGG43C6MF5BSVUEO/action/citation_signature","submit_replication":"https://pith.science/pith/KB5AWN5QHUBGG43C6MF5BSVUEO/action/replication_record"}},"created_at":"2026-08-04T00:38:04.288626+00:00","updated_at":"2026-08-04T00:38:04.288626+00:00"}