{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:XQMSJNDTGARNT27R5C4VGKRGYM","short_pith_number":"pith:XQMSJNDT","schema_version":"1.0","canonical_sha256":"bc1924b4733022d9ebf1e8b9532a26c309d928be87f7aaf1ed0c013bffc1766b","source":{"kind":"arxiv","id":"2505.19716","version":1},"attestation_state":"computed","paper":{"title":"Concise Reasoning, Big Gains: Pruning Long Reasoning Trace with Difficulty-Aware Prompting","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Bingheng Wu, Jiayi Zhang, Jingze Shi, Nan Tang, Xiaotian Lin, Yifan Wu, Yuyu Luo","submitted_at":"2025-05-26T09:04:44Z","abstract_excerpt":"Existing chain-of-thought (CoT) distillation methods can effectively transfer reasoning abilities to base models but suffer from two major limitations: excessive verbosity of reasoning traces and inadequate adaptability to problem difficulty. Long reasoning traces significantly increase inference costs, and uniform-length solutions prevent base models from learning adaptive reasoning strategies. To address these issues, we propose a difficulty-aware prompting (DAP) method to dynamically shorten reasoning traces without performance loss. In our approach, a large teacher model first judges each "},"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":"2505.19716","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2025-05-26T09:04:44Z","cross_cats_sorted":[],"title_canon_sha256":"6acebc06f1257c10f74ea08dd0d4fd4214f4d9e421d4eb83524de1e7723e4db5","abstract_canon_sha256":"ea6029477bbe735d7821791d8331a6b6f65535b183b72a385d539b4df904602b"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:09:35.076179Z","signature_b64":"aluma8q4lyjmp5i3xMrvgANS/1v3RaHRWHkCngt52KeIEmFJvuKNAtXGS3Q021xLSDLmHDq5zoh8jcRYdStPAQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"bc1924b4733022d9ebf1e8b9532a26c309d928be87f7aaf1ed0c013bffc1766b","last_reissued_at":"2026-07-05T11:09:35.075653Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:09:35.075653Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Concise Reasoning, Big Gains: Pruning Long Reasoning Trace with Difficulty-Aware Prompting","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Bingheng Wu, Jiayi Zhang, Jingze Shi, Nan Tang, Xiaotian Lin, Yifan Wu, Yuyu Luo","submitted_at":"2025-05-26T09:04:44Z","abstract_excerpt":"Existing chain-of-thought (CoT) distillation methods can effectively transfer reasoning abilities to base models but suffer from two major limitations: excessive verbosity of reasoning traces and inadequate adaptability to problem difficulty. Long reasoning traces significantly increase inference costs, and uniform-length solutions prevent base models from learning adaptive reasoning strategies. To address these issues, we propose a difficulty-aware prompting (DAP) method to dynamically shorten reasoning traces without performance loss. In our approach, a large teacher model first judges each "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2505.19716","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/2505.19716/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":"2505.19716","created_at":"2026-07-05T11:09:35.075718+00:00"},{"alias_kind":"arxiv_version","alias_value":"2505.19716v1","created_at":"2026-07-05T11:09:35.075718+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2505.19716","created_at":"2026-07-05T11:09:35.075718+00:00"},{"alias_kind":"pith_short_12","alias_value":"XQMSJNDTGARN","created_at":"2026-07-05T11:09:35.075718+00:00"},{"alias_kind":"pith_short_16","alias_value":"XQMSJNDTGARNT27R","created_at":"2026-07-05T11:09:35.075718+00:00"},{"alias_kind":"pith_short_8","alias_value":"XQMSJNDT","created_at":"2026-07-05T11:09:35.075718+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":6,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.07108","citing_title":"DyCon: Dynamic Reasoning Control via Evolving Difficulty Modeling","ref_index":36,"is_internal_anchor":false},{"citing_arxiv_id":"2504.01990","citing_title":"Advances and Challenges in Foundation Agents: From Brain-Inspired Intelligence to Evolutionary, Collaborative, and Safe Systems","ref_index":195,"is_internal_anchor":false},{"citing_arxiv_id":"2510.17586","citing_title":"DeepEye-SQL: A Software-Engineering-Inspired Text-to-SQL Framework","ref_index":55,"is_internal_anchor":false},{"citing_arxiv_id":"2605.08776","citing_title":"Reasoning Compression with Mixed-Policy Distillation","ref_index":10,"is_internal_anchor":false},{"citing_arxiv_id":"2604.19516","citing_title":"From Experience to Skill: Multi-Agent Generative Engine Optimization via Reusable Strategy Learning","ref_index":84,"is_internal_anchor":false},{"citing_arxiv_id":"2604.14847","citing_title":"TrigReason: Trigger-Based Collaboration between Small and Large Reasoning Models","ref_index":14,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/XQMSJNDTGARNT27R5C4VGKRGYM","json":"https://pith.science/pith/XQMSJNDTGARNT27R5C4VGKRGYM.json","graph_json":"https://pith.science/api/pith-number/XQMSJNDTGARNT27R5C4VGKRGYM/graph.json","events_json":"https://pith.science/api/pith-number/XQMSJNDTGARNT27R5C4VGKRGYM/events.json","paper":"https://pith.science/paper/XQMSJNDT"},"agent_actions":{"view_html":"https://pith.science/pith/XQMSJNDTGARNT27R5C4VGKRGYM","download_json":"https://pith.science/pith/XQMSJNDTGARNT27R5C4VGKRGYM.json","view_paper":"https://pith.science/paper/XQMSJNDT","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2505.19716&json=true","fetch_graph":"https://pith.science/api/pith-number/XQMSJNDTGARNT27R5C4VGKRGYM/graph.json","fetch_events":"https://pith.science/api/pith-number/XQMSJNDTGARNT27R5C4VGKRGYM/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/XQMSJNDTGARNT27R5C4VGKRGYM/action/timestamp_anchor","attest_storage":"https://pith.science/pith/XQMSJNDTGARNT27R5C4VGKRGYM/action/storage_attestation","attest_author":"https://pith.science/pith/XQMSJNDTGARNT27R5C4VGKRGYM/action/author_attestation","sign_citation":"https://pith.science/pith/XQMSJNDTGARNT27R5C4VGKRGYM/action/citation_signature","submit_replication":"https://pith.science/pith/XQMSJNDTGARNT27R5C4VGKRGYM/action/replication_record"}},"created_at":"2026-07-05T11:09:35.075718+00:00","updated_at":"2026-07-05T11:09:35.075718+00:00"}