{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:MG7EWNISFXVSVMRBSYL47Y3DIS","short_pith_number":"pith:MG7EWNIS","schema_version":"1.0","canonical_sha256":"61be4b35122deb2ab2219617cfe363448fd45e508f456d6f7d030ec64cf8cf5d","source":{"kind":"arxiv","id":"2602.13517","version":2},"attestation_state":"computed","paper":{"title":"Think Deep, Not Just Long: Measuring LLM Reasoning Effort via Deep-Thinking Tokens","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Alec Go, Blake JianHang Chen, Chao Zhao, Liqian Peng, Tian Tan, Wei-Lin Chen, Yu Meng, Ziqian Lin","submitted_at":"2026-02-13T23:07:37Z","abstract_excerpt":"Large language models (LLMs) have demonstrated impressive reasoning capabilities by scaling test-time compute via long Chain-of-Thought (CoT). However, recent findings suggest that raw token counts are unreliable proxies for reasoning quality: increased generation length does not consistently correlate with accuracy and may instead signal \"overthinking,\" leading to performance degradation. In this work, we quantify inference-time effort by identifying deep-thinking tokens -- tokens where internal predictions undergo significant revisions in deeper model layers prior to convergence. Across four"},"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":"2602.13517","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2026-02-13T23:07:37Z","cross_cats_sorted":[],"title_canon_sha256":"c17aa55cc972d1d86edb2c56a747f455e312198f961f548bdc46dd59033391f4","abstract_canon_sha256":"7901f4d579b6b0cde3ccb95ced49f761356993e7a520429010d72f63b3aa4198"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-07T02:18:34.610219Z","signature_b64":"eErEWBxQ7ZRMOAAXaaRXFsUtpgXhb9sAviwcK3qASgK/3XvWWG1832/7uqhLZ7V+Yas6mReiUqPPUfw1ixFLCQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"61be4b35122deb2ab2219617cfe363448fd45e508f456d6f7d030ec64cf8cf5d","last_reissued_at":"2026-07-07T02:18:34.608913Z","signature_status":"signed_v1","first_computed_at":"2026-07-07T02:18:34.608913Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Think Deep, Not Just Long: Measuring LLM Reasoning Effort via Deep-Thinking Tokens","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Alec Go, Blake JianHang Chen, Chao Zhao, Liqian Peng, Tian Tan, Wei-Lin Chen, Yu Meng, Ziqian Lin","submitted_at":"2026-02-13T23:07:37Z","abstract_excerpt":"Large language models (LLMs) have demonstrated impressive reasoning capabilities by scaling test-time compute via long Chain-of-Thought (CoT). However, recent findings suggest that raw token counts are unreliable proxies for reasoning quality: increased generation length does not consistently correlate with accuracy and may instead signal \"overthinking,\" leading to performance degradation. In this work, we quantify inference-time effort by identifying deep-thinking tokens -- tokens where internal predictions undergo significant revisions in deeper model layers prior to convergence. Across four"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2602.13517","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/2602.13517/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":"2602.13517","created_at":"2026-07-07T02:18:34.609040+00:00"},{"alias_kind":"arxiv_version","alias_value":"2602.13517v2","created_at":"2026-07-07T02:18:34.609040+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2602.13517","created_at":"2026-07-07T02:18:34.609040+00:00"},{"alias_kind":"pith_short_12","alias_value":"MG7EWNISFXVS","created_at":"2026-07-07T02:18:34.609040+00:00"},{"alias_kind":"pith_short_16","alias_value":"MG7EWNISFXVSVMRB","created_at":"2026-07-07T02:18:34.609040+00:00"},{"alias_kind":"pith_short_8","alias_value":"MG7EWNIS","created_at":"2026-07-07T02:18:34.609040+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":11,"internal_anchor_count":11,"sample":[{"citing_arxiv_id":"2606.21972","citing_title":"Human vs Machine Mathematical Difficulty on Project Euler: An Experimental Analysis","ref_index":2,"is_internal_anchor":true},{"citing_arxiv_id":"2606.08755","citing_title":"Co-Evolving Skill Generation and Policy Optimization","ref_index":9,"is_internal_anchor":true},{"citing_arxiv_id":"2605.28006","citing_title":"Integrated and Cross-Architecture Interpretation of LLM Reasoning","ref_index":5,"is_internal_anchor":true},{"citing_arxiv_id":"2605.30666","citing_title":"The Tutoring Effectiveness Index: Predicting LLM Math Tutor Quality from Four Conversation Signals","ref_index":3,"is_internal_anchor":true},{"citing_arxiv_id":"2605.22873","citing_title":"When Do LLMs Reason? A Dynamical Systems View via Entropy Phase Transitions","ref_index":36,"is_internal_anchor":true},{"citing_arxiv_id":"2605.14457","citing_title":"Stateful Reasoning via Insight Replay","ref_index":6,"is_internal_anchor":true},{"citing_arxiv_id":"2605.15764","citing_title":"GRASP: Learning to Ground Social Reasoning in Multi-Person Non-Verbal Interactions","ref_index":11,"is_internal_anchor":true},{"citing_arxiv_id":"2605.16938","citing_title":"Effort as Ceiling, Not Dial: Reasoning Budget Does Not Modulate Cognitive Cost Alignment Between Humans and Large Reasoning Models","ref_index":5,"is_internal_anchor":true},{"citing_arxiv_id":"2605.14457","citing_title":"Stateful Reasoning via Insight Replay","ref_index":6,"is_internal_anchor":true},{"citing_arxiv_id":"2605.01853","citing_title":"Spatiotemporal Hidden-State Dynamics as a Signature of Internal Reasoning in Large Language Models","ref_index":18,"is_internal_anchor":true},{"citing_arxiv_id":"2605.01694","citing_title":"Latent State Design for World Models under Sufficiency Constraints","ref_index":11,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/MG7EWNISFXVSVMRBSYL47Y3DIS","json":"https://pith.science/pith/MG7EWNISFXVSVMRBSYL47Y3DIS.json","graph_json":"https://pith.science/api/pith-number/MG7EWNISFXVSVMRBSYL47Y3DIS/graph.json","events_json":"https://pith.science/api/pith-number/MG7EWNISFXVSVMRBSYL47Y3DIS/events.json","paper":"https://pith.science/paper/MG7EWNIS"},"agent_actions":{"view_html":"https://pith.science/pith/MG7EWNISFXVSVMRBSYL47Y3DIS","download_json":"https://pith.science/pith/MG7EWNISFXVSVMRBSYL47Y3DIS.json","view_paper":"https://pith.science/paper/MG7EWNIS","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2602.13517&json=true","fetch_graph":"https://pith.science/api/pith-number/MG7EWNISFXVSVMRBSYL47Y3DIS/graph.json","fetch_events":"https://pith.science/api/pith-number/MG7EWNISFXVSVMRBSYL47Y3DIS/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/MG7EWNISFXVSVMRBSYL47Y3DIS/action/timestamp_anchor","attest_storage":"https://pith.science/pith/MG7EWNISFXVSVMRBSYL47Y3DIS/action/storage_attestation","attest_author":"https://pith.science/pith/MG7EWNISFXVSVMRBSYL47Y3DIS/action/author_attestation","sign_citation":"https://pith.science/pith/MG7EWNISFXVSVMRBSYL47Y3DIS/action/citation_signature","submit_replication":"https://pith.science/pith/MG7EWNISFXVSVMRBSYL47Y3DIS/action/replication_record"}},"created_at":"2026-07-07T02:18:34.609040+00:00","updated_at":"2026-07-07T02:18:34.609040+00:00"}