{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:XGJLYJ2CDLK32FYWC5ZAJEO2UI","short_pith_number":"pith:XGJLYJ2C","schema_version":"1.0","canonical_sha256":"b992bc27421ad5bd171617720491daa228cac0096e2f18b6b3ac3ac5fb3f59bf","source":{"kind":"arxiv","id":"2504.03234","version":2},"attestation_state":"computed","paper":{"title":"Think When You Need: Self-Adaptive Chain-of-Thought Learning","license":"http://creativecommons.org/licenses/by-sa/4.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"Junjie Yang, Ke Lin, Xing Yu","submitted_at":"2025-04-04T07:34:01Z","abstract_excerpt":"Chain of Thought (CoT) reasoning enhances language models' performance but often leads to inefficient \"overthinking\" on simple problems. We identify that existing approaches directly penalizing reasoning length fail to account for varying problem complexity. Our approach constructs rewards through length and quality comparisons, guided by theoretical assumptions that jointly enhance solution correctness with conciseness. Moreover, we further demonstrate our method to fuzzy tasks where ground truth is unavailable. Experiments across multiple reasoning benchmarks demonstrate that our method main"},"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":"2504.03234","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.CL","submitted_at":"2025-04-04T07:34:01Z","cross_cats_sorted":["cs.AI","cs.LG"],"title_canon_sha256":"1a21f949c2f5b923aee35baa334b7cf3a8ad731fcc755f6c1e676309f517e0bb","abstract_canon_sha256":"5ae5394043c869c76babb9d6d677a7015635f735ba211ebb48dab0df6f786faf"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:06:30.159712Z","signature_b64":"f9BXO6xq4zSNm3djGa1qGwPTcNjWy18FBvswAbidq9KvfdXPgL+wDsdB9bGBXWHHm6QiyPpfvsvZ4lSOuvHAAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"b992bc27421ad5bd171617720491daa228cac0096e2f18b6b3ac3ac5fb3f59bf","last_reissued_at":"2026-07-05T11:06:30.159133Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:06:30.159133Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Think When You Need: Self-Adaptive Chain-of-Thought Learning","license":"http://creativecommons.org/licenses/by-sa/4.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"Junjie Yang, Ke Lin, Xing Yu","submitted_at":"2025-04-04T07:34:01Z","abstract_excerpt":"Chain of Thought (CoT) reasoning enhances language models' performance but often leads to inefficient \"overthinking\" on simple problems. We identify that existing approaches directly penalizing reasoning length fail to account for varying problem complexity. Our approach constructs rewards through length and quality comparisons, guided by theoretical assumptions that jointly enhance solution correctness with conciseness. Moreover, we further demonstrate our method to fuzzy tasks where ground truth is unavailable. Experiments across multiple reasoning benchmarks demonstrate that our method main"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2504.03234","kind":"arxiv","version":2},"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/2504.03234/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":"2504.03234","created_at":"2026-07-05T11:06:30.159185+00:00"},{"alias_kind":"arxiv_version","alias_value":"2504.03234v2","created_at":"2026-07-05T11:06:30.159185+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2504.03234","created_at":"2026-07-05T11:06:30.159185+00:00"},{"alias_kind":"pith_short_12","alias_value":"XGJLYJ2CDLK3","created_at":"2026-07-05T11:06:30.159185+00:00"},{"alias_kind":"pith_short_16","alias_value":"XGJLYJ2CDLK32FYW","created_at":"2026-07-05T11:06:30.159185+00:00"},{"alias_kind":"pith_short_8","alias_value":"XGJLYJ2C","created_at":"2026-07-05T11:06:30.159185+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":5,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.05988","citing_title":"Compress-Distill: Reasoning Trace Compression for Efficient Knowledge Distillation","ref_index":41,"is_internal_anchor":false},{"citing_arxiv_id":"2605.30832","citing_title":"SLAT: Segment-Level Adaptive Trimming for Efficient CoT Reasoning","ref_index":27,"is_internal_anchor":false},{"citing_arxiv_id":"2602.06475","citing_title":"Towards Generalizable Reasoning: Group Causal Counterfactual Policy Optimization for LLM Reasoning","ref_index":15,"is_internal_anchor":false},{"citing_arxiv_id":"2605.13165","citing_title":"STOP: Structured On-Policy Pruning of Long-Form Reasoning in Low-Data Regimes","ref_index":74,"is_internal_anchor":false},{"citing_arxiv_id":"2604.14847","citing_title":"TrigReason: Trigger-Based Collaboration between Small and Large Reasoning Models","ref_index":17,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/XGJLYJ2CDLK32FYWC5ZAJEO2UI","json":"https://pith.science/pith/XGJLYJ2CDLK32FYWC5ZAJEO2UI.json","graph_json":"https://pith.science/api/pith-number/XGJLYJ2CDLK32FYWC5ZAJEO2UI/graph.json","events_json":"https://pith.science/api/pith-number/XGJLYJ2CDLK32FYWC5ZAJEO2UI/events.json","paper":"https://pith.science/paper/XGJLYJ2C"},"agent_actions":{"view_html":"https://pith.science/pith/XGJLYJ2CDLK32FYWC5ZAJEO2UI","download_json":"https://pith.science/pith/XGJLYJ2CDLK32FYWC5ZAJEO2UI.json","view_paper":"https://pith.science/paper/XGJLYJ2C","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2504.03234&json=true","fetch_graph":"https://pith.science/api/pith-number/XGJLYJ2CDLK32FYWC5ZAJEO2UI/graph.json","fetch_events":"https://pith.science/api/pith-number/XGJLYJ2CDLK32FYWC5ZAJEO2UI/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/XGJLYJ2CDLK32FYWC5ZAJEO2UI/action/timestamp_anchor","attest_storage":"https://pith.science/pith/XGJLYJ2CDLK32FYWC5ZAJEO2UI/action/storage_attestation","attest_author":"https://pith.science/pith/XGJLYJ2CDLK32FYWC5ZAJEO2UI/action/author_attestation","sign_citation":"https://pith.science/pith/XGJLYJ2CDLK32FYWC5ZAJEO2UI/action/citation_signature","submit_replication":"https://pith.science/pith/XGJLYJ2CDLK32FYWC5ZAJEO2UI/action/replication_record"}},"created_at":"2026-07-05T11:06:30.159185+00:00","updated_at":"2026-07-05T11:06:30.159185+00:00"}