{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:6N5J3XZUQYZR4VJX4R4LL7XHDX","short_pith_number":"pith:6N5J3XZU","schema_version":"1.0","canonical_sha256":"f37a9ddf3486331e5537e478b5fee71ddac5d631c54ef4cd51934bc9bebafb30","source":{"kind":"arxiv","id":"2504.13367","version":1},"attestation_state":"computed","paper":{"title":"THOUGHTTERMINATOR: Benchmarking, Calibrating, and Mitigating Overthinking in Reasoning Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Michael Saxon, Wenyue Hua, William Yang Wang, Xiao Pu","submitted_at":"2025-04-17T22:16:30Z","abstract_excerpt":"Reasoning models have demonstrated impressive performance on difficult tasks that traditional language models struggle at. However, many are plagued with the problem of overthinking--generating large amounts of unnecessary tokens which don't improve accuracy on a question. We introduce approximate measures of problem-level difficulty and demonstrate that a clear relationship between problem difficulty and optimal token spend exists, and evaluate how well calibrated a variety of reasoning models are in terms of efficiently allocating the optimal token count. We find that in general, reasoning m"},"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":"2504.13367","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2025-04-17T22:16:30Z","cross_cats_sorted":[],"title_canon_sha256":"81e820451877f4f9eb1b4302a800b548b72c62b8931d5612f261af358e5a1514","abstract_canon_sha256":"078eea4e6461dff53ff0b884804390eb78c82a01c6ce1c5d885e163909ae044e"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:50:53.722858Z","signature_b64":"ZQr3qxUKvIzAMxyKprfmLOPH9SOXAu9EYmVHEkIuv7BzBgZf6EzFfcjJv+EhuzCMGl7XWi3ZOrchNB5HCZZUBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"f37a9ddf3486331e5537e478b5fee71ddac5d631c54ef4cd51934bc9bebafb30","last_reissued_at":"2026-07-05T10:50:53.722291Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:50:53.722291Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"THOUGHTTERMINATOR: Benchmarking, Calibrating, and Mitigating Overthinking in Reasoning Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Michael Saxon, Wenyue Hua, William Yang Wang, Xiao Pu","submitted_at":"2025-04-17T22:16:30Z","abstract_excerpt":"Reasoning models have demonstrated impressive performance on difficult tasks that traditional language models struggle at. However, many are plagued with the problem of overthinking--generating large amounts of unnecessary tokens which don't improve accuracy on a question. We introduce approximate measures of problem-level difficulty and demonstrate that a clear relationship between problem difficulty and optimal token spend exists, and evaluate how well calibrated a variety of reasoning models are in terms of efficiently allocating the optimal token count. We find that in general, reasoning m"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2504.13367","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/2504.13367/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":"2504.13367","created_at":"2026-07-05T10:50:53.722359+00:00"},{"alias_kind":"arxiv_version","alias_value":"2504.13367v1","created_at":"2026-07-05T10:50:53.722359+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2504.13367","created_at":"2026-07-05T10:50:53.722359+00:00"},{"alias_kind":"pith_short_12","alias_value":"6N5J3XZUQYZR","created_at":"2026-07-05T10:50:53.722359+00:00"},{"alias_kind":"pith_short_16","alias_value":"6N5J3XZUQYZR4VJX","created_at":"2026-07-05T10:50:53.722359+00:00"},{"alias_kind":"pith_short_8","alias_value":"6N5J3XZU","created_at":"2026-07-05T10:50:53.722359+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.17890","citing_title":"Dynamic Rollout Editing for Reducing Overthinking in RL-Trained Reasoning Models","ref_index":48,"is_internal_anchor":false},{"citing_arxiv_id":"2507.04023","citing_title":"Do LLMs Overthink Basic Math Reasoning? Benchmarking the Accuracy-Efficiency Tradeoff in Language Models","ref_index":18,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/6N5J3XZUQYZR4VJX4R4LL7XHDX","json":"https://pith.science/pith/6N5J3XZUQYZR4VJX4R4LL7XHDX.json","graph_json":"https://pith.science/api/pith-number/6N5J3XZUQYZR4VJX4R4LL7XHDX/graph.json","events_json":"https://pith.science/api/pith-number/6N5J3XZUQYZR4VJX4R4LL7XHDX/events.json","paper":"https://pith.science/paper/6N5J3XZU"},"agent_actions":{"view_html":"https://pith.science/pith/6N5J3XZUQYZR4VJX4R4LL7XHDX","download_json":"https://pith.science/pith/6N5J3XZUQYZR4VJX4R4LL7XHDX.json","view_paper":"https://pith.science/paper/6N5J3XZU","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2504.13367&json=true","fetch_graph":"https://pith.science/api/pith-number/6N5J3XZUQYZR4VJX4R4LL7XHDX/graph.json","fetch_events":"https://pith.science/api/pith-number/6N5J3XZUQYZR4VJX4R4LL7XHDX/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/6N5J3XZUQYZR4VJX4R4LL7XHDX/action/timestamp_anchor","attest_storage":"https://pith.science/pith/6N5J3XZUQYZR4VJX4R4LL7XHDX/action/storage_attestation","attest_author":"https://pith.science/pith/6N5J3XZUQYZR4VJX4R4LL7XHDX/action/author_attestation","sign_citation":"https://pith.science/pith/6N5J3XZUQYZR4VJX4R4LL7XHDX/action/citation_signature","submit_replication":"https://pith.science/pith/6N5J3XZUQYZR4VJX4R4LL7XHDX/action/replication_record"}},"created_at":"2026-07-05T10:50:53.722359+00:00","updated_at":"2026-07-05T10:50:53.722359+00:00"}