{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:A2LFHJZRFPLJCK2UDBKQ4WPMJW","short_pith_number":"pith:A2LFHJZR","schema_version":"1.0","canonical_sha256":"069653a7312bd6912b5418550e59ec4d881d0a8653887ec75a26e0bcf4291ea3","source":{"kind":"arxiv","id":"2506.06923","version":1},"attestation_state":"computed","paper":{"title":"Boosting LLM Reasoning via Spontaneous Self-Correction","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Chen Zhu, Di Jin, Han Fang, Liang Tan, Sarath Chandar, Sinong Wang, Tengyu Xu, Xuewei Wang, Xutong Zhao, Yen-Ting, Yun He, Zhengxing Chen, Zhuokai Zhao, Zishun Yu","submitted_at":"2025-06-07T21:23:00Z","abstract_excerpt":"While large language models (LLMs) have demonstrated remarkable success on a broad range of tasks, math reasoning remains a challenging one. One of the approaches for improving math reasoning is self-correction, which designs self-improving loops to let the model correct its own mistakes. However, existing self-correction approaches treat corrections as standalone post-generation refinements, relying on extra prompt and system designs to elicit self-corrections, instead of performing real-time, spontaneous self-corrections in a single pass. To address this, we propose SPOC, a spontaneous self-"},"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":"2506.06923","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2025-06-07T21:23:00Z","cross_cats_sorted":[],"title_canon_sha256":"90b336af4cce8154df370063cc4d096a5b01078058a493630ae01ce91fbd2669","abstract_canon_sha256":"a2d45dba6269a201b0414cb02ee51bde85728f4e90b2fe1cccc12c26d7648ed6"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:17:58.273662Z","signature_b64":"CZTozCpS3RpPeHLSfidg9aLsRumiPSOgcuH3hzHI5M/6mStxxDTzfQBtYJntlXK3qhQZOmUFc2LeMjh1ecZHAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"069653a7312bd6912b5418550e59ec4d881d0a8653887ec75a26e0bcf4291ea3","last_reissued_at":"2026-07-05T11:17:58.273158Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:17:58.273158Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Boosting LLM Reasoning via Spontaneous Self-Correction","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Chen Zhu, Di Jin, Han Fang, Liang Tan, Sarath Chandar, Sinong Wang, Tengyu Xu, Xuewei Wang, Xutong Zhao, Yen-Ting, Yun He, Zhengxing Chen, Zhuokai Zhao, Zishun Yu","submitted_at":"2025-06-07T21:23:00Z","abstract_excerpt":"While large language models (LLMs) have demonstrated remarkable success on a broad range of tasks, math reasoning remains a challenging one. One of the approaches for improving math reasoning is self-correction, which designs self-improving loops to let the model correct its own mistakes. However, existing self-correction approaches treat corrections as standalone post-generation refinements, relying on extra prompt and system designs to elicit self-corrections, instead of performing real-time, spontaneous self-corrections in a single pass. To address this, we propose SPOC, a spontaneous self-"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2506.06923","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/2506.06923/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":"2506.06923","created_at":"2026-07-05T11:17:58.273218+00:00"},{"alias_kind":"arxiv_version","alias_value":"2506.06923v1","created_at":"2026-07-05T11:17:58.273218+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2506.06923","created_at":"2026-07-05T11:17:58.273218+00:00"},{"alias_kind":"pith_short_12","alias_value":"A2LFHJZRFPLJ","created_at":"2026-07-05T11:17:58.273218+00:00"},{"alias_kind":"pith_short_16","alias_value":"A2LFHJZRFPLJCK2U","created_at":"2026-07-05T11:17:58.273218+00:00"},{"alias_kind":"pith_short_8","alias_value":"A2LFHJZR","created_at":"2026-07-05T11:17:58.273218+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":5,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.13316","citing_title":"ReSum: Synergizing LLM Reasoning and Summarization with Reinforcement Learning","ref_index":74,"is_internal_anchor":false},{"citing_arxiv_id":"2606.29425","citing_title":"Mixture of Debaters: Learn to Debate at Architectural Level in Multi-Agent Reasoning","ref_index":57,"is_internal_anchor":false},{"citing_arxiv_id":"2606.05976","citing_title":"The Self-Correction Illusion: Role Relabeling Gates Explicit Error Flagging in Large Language Models","ref_index":50,"is_internal_anchor":false},{"citing_arxiv_id":"2601.05106","citing_title":"Token-Level LLM Collaboration via FusionRoute","ref_index":30,"is_internal_anchor":false},{"citing_arxiv_id":"2509.18847","citing_title":"Failure Makes the Agent Stronger: Enhancing Accuracy through Structured Reflection for Reliable Tool Interactions","ref_index":29,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/A2LFHJZRFPLJCK2UDBKQ4WPMJW","json":"https://pith.science/pith/A2LFHJZRFPLJCK2UDBKQ4WPMJW.json","graph_json":"https://pith.science/api/pith-number/A2LFHJZRFPLJCK2UDBKQ4WPMJW/graph.json","events_json":"https://pith.science/api/pith-number/A2LFHJZRFPLJCK2UDBKQ4WPMJW/events.json","paper":"https://pith.science/paper/A2LFHJZR"},"agent_actions":{"view_html":"https://pith.science/pith/A2LFHJZRFPLJCK2UDBKQ4WPMJW","download_json":"https://pith.science/pith/A2LFHJZRFPLJCK2UDBKQ4WPMJW.json","view_paper":"https://pith.science/paper/A2LFHJZR","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2506.06923&json=true","fetch_graph":"https://pith.science/api/pith-number/A2LFHJZRFPLJCK2UDBKQ4WPMJW/graph.json","fetch_events":"https://pith.science/api/pith-number/A2LFHJZRFPLJCK2UDBKQ4WPMJW/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/A2LFHJZRFPLJCK2UDBKQ4WPMJW/action/timestamp_anchor","attest_storage":"https://pith.science/pith/A2LFHJZRFPLJCK2UDBKQ4WPMJW/action/storage_attestation","attest_author":"https://pith.science/pith/A2LFHJZRFPLJCK2UDBKQ4WPMJW/action/author_attestation","sign_citation":"https://pith.science/pith/A2LFHJZRFPLJCK2UDBKQ4WPMJW/action/citation_signature","submit_replication":"https://pith.science/pith/A2LFHJZRFPLJCK2UDBKQ4WPMJW/action/replication_record"}},"created_at":"2026-07-05T11:17:58.273218+00:00","updated_at":"2026-07-05T11:17:58.273218+00:00"}