{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:M2FKHBTFYK3ZUSA7X3NRBBJSPD","short_pith_number":"pith:M2FKHBTF","schema_version":"1.0","canonical_sha256":"668aa38665c2b79a481fbedb10853278da651ad6ce7220abeba1bfc2d63a6b35","source":{"kind":"arxiv","id":"2505.18237","version":1},"attestation_state":"computed","paper":{"title":"Think or Not? Exploring Thinking Efficiency in Large Reasoning Models via an Information-Theoretic Lens","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.IT","math.IT"],"primary_cat":"cs.CL","authors_text":"Jinlin Li, Xian Wu, Xiao Zhou, Xixian Yong, Yefeng Zheng, Yingying Zhang","submitted_at":"2025-05-23T13:38:56Z","abstract_excerpt":"The recent rise of Large Reasoning Models (LRMs) has significantly improved multi-step reasoning performance, but often at the cost of generating excessively long reasoning chains. This paper revisits the efficiency of such reasoning processes through an information-theoretic lens, revealing a fundamental trade-off between reasoning length and semantic efficiency. We propose two metrics, InfoBias and InfoGain, to quantify divergence from ideal reasoning paths and stepwise information contribution, respectively. Empirical analyses show that longer reasoning chains tend to exhibit higher informa"},"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":"2505.18237","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2025-05-23T13:38:56Z","cross_cats_sorted":["cs.AI","cs.IT","math.IT"],"title_canon_sha256":"39ac34446b9e76f936ef4e6b461e7d38b3f0b7ab20a59f6dea55c3aead318cf4","abstract_canon_sha256":"8600faa758c9953953346f94a75bc3859b5648a2f0dc1c47803185cedf5e1c52"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:08:43.320490Z","signature_b64":"L2RMXIBM8kahsdf37dIbleS7qE19DHpxwG+b693fpyZiWaq9+73M+zG4vfBBryeaQvF8hFEJvb+1g+I/07ihCw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"668aa38665c2b79a481fbedb10853278da651ad6ce7220abeba1bfc2d63a6b35","last_reissued_at":"2026-07-05T11:08:43.320006Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:08:43.320006Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Think or Not? Exploring Thinking Efficiency in Large Reasoning Models via an Information-Theoretic Lens","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.IT","math.IT"],"primary_cat":"cs.CL","authors_text":"Jinlin Li, Xian Wu, Xiao Zhou, Xixian Yong, Yefeng Zheng, Yingying Zhang","submitted_at":"2025-05-23T13:38:56Z","abstract_excerpt":"The recent rise of Large Reasoning Models (LRMs) has significantly improved multi-step reasoning performance, but often at the cost of generating excessively long reasoning chains. This paper revisits the efficiency of such reasoning processes through an information-theoretic lens, revealing a fundamental trade-off between reasoning length and semantic efficiency. We propose two metrics, InfoBias and InfoGain, to quantify divergence from ideal reasoning paths and stepwise information contribution, respectively. Empirical analyses show that longer reasoning chains tend to exhibit higher informa"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2505.18237","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/2505.18237/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":"2505.18237","created_at":"2026-07-05T11:08:43.320065+00:00"},{"alias_kind":"arxiv_version","alias_value":"2505.18237v1","created_at":"2026-07-05T11:08:43.320065+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2505.18237","created_at":"2026-07-05T11:08:43.320065+00:00"},{"alias_kind":"pith_short_12","alias_value":"M2FKHBTFYK3Z","created_at":"2026-07-05T11:08:43.320065+00:00"},{"alias_kind":"pith_short_16","alias_value":"M2FKHBTFYK3ZUSA7","created_at":"2026-07-05T11:08:43.320065+00:00"},{"alias_kind":"pith_short_8","alias_value":"M2FKHBTF","created_at":"2026-07-05T11:08:43.320065+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":6,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.03102","citing_title":"Small RL Controller, Large Language Model: RL-Guided Adaptive Sampling for Test-Time Scaling","ref_index":92,"is_internal_anchor":false},{"citing_arxiv_id":"2509.26522","citing_title":"Entropy After </Think> for reasoning model early exiting","ref_index":22,"is_internal_anchor":false},{"citing_arxiv_id":"2605.14062","citing_title":"Know When To Fold 'Em: Token-Efficient LLM Synthetic Data Generation via Multi-Stage In-Flight Rejection","ref_index":43,"is_internal_anchor":false},{"citing_arxiv_id":"2605.08083","citing_title":"LLMs Improving LLMs: Agentic Discovery for Test-Time Scaling","ref_index":33,"is_internal_anchor":false},{"citing_arxiv_id":"2605.08083","citing_title":"LLMs Improving LLMs: Agentic Discovery for Test-Time Scaling","ref_index":33,"is_internal_anchor":false},{"citing_arxiv_id":"2604.14528","citing_title":"Dissecting Failure Dynamics in Large Language Model Reasoning","ref_index":2,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/M2FKHBTFYK3ZUSA7X3NRBBJSPD","json":"https://pith.science/pith/M2FKHBTFYK3ZUSA7X3NRBBJSPD.json","graph_json":"https://pith.science/api/pith-number/M2FKHBTFYK3ZUSA7X3NRBBJSPD/graph.json","events_json":"https://pith.science/api/pith-number/M2FKHBTFYK3ZUSA7X3NRBBJSPD/events.json","paper":"https://pith.science/paper/M2FKHBTF"},"agent_actions":{"view_html":"https://pith.science/pith/M2FKHBTFYK3ZUSA7X3NRBBJSPD","download_json":"https://pith.science/pith/M2FKHBTFYK3ZUSA7X3NRBBJSPD.json","view_paper":"https://pith.science/paper/M2FKHBTF","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2505.18237&json=true","fetch_graph":"https://pith.science/api/pith-number/M2FKHBTFYK3ZUSA7X3NRBBJSPD/graph.json","fetch_events":"https://pith.science/api/pith-number/M2FKHBTFYK3ZUSA7X3NRBBJSPD/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/M2FKHBTFYK3ZUSA7X3NRBBJSPD/action/timestamp_anchor","attest_storage":"https://pith.science/pith/M2FKHBTFYK3ZUSA7X3NRBBJSPD/action/storage_attestation","attest_author":"https://pith.science/pith/M2FKHBTFYK3ZUSA7X3NRBBJSPD/action/author_attestation","sign_citation":"https://pith.science/pith/M2FKHBTFYK3ZUSA7X3NRBBJSPD/action/citation_signature","submit_replication":"https://pith.science/pith/M2FKHBTFYK3ZUSA7X3NRBBJSPD/action/replication_record"}},"created_at":"2026-07-05T11:08:43.320065+00:00","updated_at":"2026-07-05T11:08:43.320065+00:00"}