{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:XKQ4OSRHAI6LCUXJ3DBNWVQDBA","short_pith_number":"pith:XKQ4OSRH","schema_version":"1.0","canonical_sha256":"baa1c74a27023cb152e9d8c2db560308152ebff135afce93ebfb54324bb92592","source":{"kind":"arxiv","id":"2507.04348","version":2},"attestation_state":"computed","paper":{"title":"SmartThinker: Learning to Compress and Preserve Reasoning by Step-Level Length Control","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.AI","authors_text":"Jie Liu, Xiao Ling, Xingyang He","submitted_at":"2025-07-06T11:21:47Z","abstract_excerpt":"Large reasoning models (LRMs) have exhibited remarkable reasoning capabilities through inference-time scaling, but this progress has also introduced considerable redundancy and inefficiency into their reasoning processes, resulting in substantial computational waste. Previous work has attempted to mitigate this issue by penalizing the overall length of generated samples during reinforcement learning (RL), with the goal of encouraging a more concise chains of thought. However, we observe that such global length penalty often lead to excessive compression of critical reasoning steps while preser"},"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":"2507.04348","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AI","submitted_at":"2025-07-06T11:21:47Z","cross_cats_sorted":["cs.CL"],"title_canon_sha256":"3985814c7f8cb4015eed10f007f4f44be3b673f1270aecb05d2fa53aa35b6623","abstract_canon_sha256":"274acb14be2a91a451d6e3eeb8cf90092ecda7e63feed28eab206f732e0cf639"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:38:52.016535Z","signature_b64":"GKjGgiBXKRYorH3LYuasnWfpC0Ncn48VsgU3SrilXtUrD60JPMC1C4zP3Ojx8w/9kTtPmj6I6gJfaKfgvup0DQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"baa1c74a27023cb152e9d8c2db560308152ebff135afce93ebfb54324bb92592","last_reissued_at":"2026-07-05T11:38:52.016059Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:38:52.016059Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"SmartThinker: Learning to Compress and Preserve Reasoning by Step-Level Length Control","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.AI","authors_text":"Jie Liu, Xiao Ling, Xingyang He","submitted_at":"2025-07-06T11:21:47Z","abstract_excerpt":"Large reasoning models (LRMs) have exhibited remarkable reasoning capabilities through inference-time scaling, but this progress has also introduced considerable redundancy and inefficiency into their reasoning processes, resulting in substantial computational waste. Previous work has attempted to mitigate this issue by penalizing the overall length of generated samples during reinforcement learning (RL), with the goal of encouraging a more concise chains of thought. However, we observe that such global length penalty often lead to excessive compression of critical reasoning steps while preser"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2507.04348","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/2507.04348/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":"2507.04348","created_at":"2026-07-05T11:38:52.016111+00:00"},{"alias_kind":"arxiv_version","alias_value":"2507.04348v2","created_at":"2026-07-05T11:38:52.016111+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2507.04348","created_at":"2026-07-05T11:38:52.016111+00:00"},{"alias_kind":"pith_short_12","alias_value":"XKQ4OSRHAI6L","created_at":"2026-07-05T11:38:52.016111+00:00"},{"alias_kind":"pith_short_16","alias_value":"XKQ4OSRHAI6LCUXJ","created_at":"2026-07-05T11:38:52.016111+00:00"},{"alias_kind":"pith_short_8","alias_value":"XKQ4OSRH","created_at":"2026-07-05T11:38:52.016111+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2503.09567","citing_title":"Towards Reasoning Era: A Survey of Long Chain-of-Thought for Reasoning Large Language Models","ref_index":254,"is_internal_anchor":false},{"citing_arxiv_id":"2605.09806","citing_title":"LEAD: Length-Efficient Adaptive and Dynamic Reasoning for Large Language Models","ref_index":17,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/XKQ4OSRHAI6LCUXJ3DBNWVQDBA","json":"https://pith.science/pith/XKQ4OSRHAI6LCUXJ3DBNWVQDBA.json","graph_json":"https://pith.science/api/pith-number/XKQ4OSRHAI6LCUXJ3DBNWVQDBA/graph.json","events_json":"https://pith.science/api/pith-number/XKQ4OSRHAI6LCUXJ3DBNWVQDBA/events.json","paper":"https://pith.science/paper/XKQ4OSRH"},"agent_actions":{"view_html":"https://pith.science/pith/XKQ4OSRHAI6LCUXJ3DBNWVQDBA","download_json":"https://pith.science/pith/XKQ4OSRHAI6LCUXJ3DBNWVQDBA.json","view_paper":"https://pith.science/paper/XKQ4OSRH","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2507.04348&json=true","fetch_graph":"https://pith.science/api/pith-number/XKQ4OSRHAI6LCUXJ3DBNWVQDBA/graph.json","fetch_events":"https://pith.science/api/pith-number/XKQ4OSRHAI6LCUXJ3DBNWVQDBA/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/XKQ4OSRHAI6LCUXJ3DBNWVQDBA/action/timestamp_anchor","attest_storage":"https://pith.science/pith/XKQ4OSRHAI6LCUXJ3DBNWVQDBA/action/storage_attestation","attest_author":"https://pith.science/pith/XKQ4OSRHAI6LCUXJ3DBNWVQDBA/action/author_attestation","sign_citation":"https://pith.science/pith/XKQ4OSRHAI6LCUXJ3DBNWVQDBA/action/citation_signature","submit_replication":"https://pith.science/pith/XKQ4OSRHAI6LCUXJ3DBNWVQDBA/action/replication_record"}},"created_at":"2026-07-05T11:38:52.016111+00:00","updated_at":"2026-07-05T11:38:52.016111+00:00"}