{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:THZM232D6HNPCOROAYDBJJHYDC","short_pith_number":"pith:THZM232D","schema_version":"1.0","canonical_sha256":"99f2cd6f43f1daf13a2e060614a4f818bbccdd55c2ddabdd52a334ce514a0ba8","source":{"kind":"arxiv","id":"2508.09883","version":2},"attestation_state":"computed","paper":{"title":"Beyond Scaling Law: A Data-Efficient Distillation Framework for Reasoning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Dengfeng Liu, Huang Liu, Huiyang Li, Jin Yang, Ji Xie, Jucai Zhai, Jun Wang, Kailai Zhang, Ninglun Gu, Qiaobo Hao, Xiaoguang Jiang, Xiaojun Wu, Yelun Bao, Zhiguo Yang","submitted_at":"2025-08-13T15:32:25Z","abstract_excerpt":"Large language models (LLMs) demonstrate remarkable reasoning capabilities in tasks such as algorithmic coding and mathematical problem-solving. Recent methods have improved reasoning through expanded corpus and multistage training combining reinforcement learning and supervised fine-tuning. Although some methods suggest that small but targeted dataset can incentivize reasoning via only distillation, a reasoning scaling laws is still taking shape, increasing computational costs. To address this, we propose a data-efficient distillation framework (DED) that optimizes the Pareto frontier of reas"},"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":"2508.09883","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-08-13T15:32:25Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"e960339892690d540bec690737887cc0b287d4c425cc85e637b78e887781c9fc","abstract_canon_sha256":"d6f595bd9bdba8096fea18678dddfd9655fe6dd6485faf3cfe90fed979d729e4"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-30T02:16:53.655103Z","signature_b64":"BLqMpJby8t+QNiKDSKmKuxe5Bj1vFgHcik3EqeDP9oCey7CzhvqmQqHZBbSf67T86bEL4tz0cefarj4pp5v9Bg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"99f2cd6f43f1daf13a2e060614a4f818bbccdd55c2ddabdd52a334ce514a0ba8","last_reissued_at":"2026-06-30T02:16:53.654462Z","signature_status":"signed_v1","first_computed_at":"2026-06-30T02:16:53.654462Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Beyond Scaling Law: A Data-Efficient Distillation Framework for Reasoning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Dengfeng Liu, Huang Liu, Huiyang Li, Jin Yang, Ji Xie, Jucai Zhai, Jun Wang, Kailai Zhang, Ninglun Gu, Qiaobo Hao, Xiaoguang Jiang, Xiaojun Wu, Yelun Bao, Zhiguo Yang","submitted_at":"2025-08-13T15:32:25Z","abstract_excerpt":"Large language models (LLMs) demonstrate remarkable reasoning capabilities in tasks such as algorithmic coding and mathematical problem-solving. Recent methods have improved reasoning through expanded corpus and multistage training combining reinforcement learning and supervised fine-tuning. Although some methods suggest that small but targeted dataset can incentivize reasoning via only distillation, a reasoning scaling laws is still taking shape, increasing computational costs. To address this, we propose a data-efficient distillation framework (DED) that optimizes the Pareto frontier of reas"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2508.09883","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/2508.09883/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":"2508.09883","created_at":"2026-06-30T02:16:53.654539+00:00"},{"alias_kind":"arxiv_version","alias_value":"2508.09883v2","created_at":"2026-06-30T02:16:53.654539+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2508.09883","created_at":"2026-06-30T02:16:53.654539+00:00"},{"alias_kind":"pith_short_12","alias_value":"THZM232D6HNP","created_at":"2026-06-30T02:16:53.654539+00:00"},{"alias_kind":"pith_short_16","alias_value":"THZM232D6HNPCORO","created_at":"2026-06-30T02:16:53.654539+00:00"},{"alias_kind":"pith_short_8","alias_value":"THZM232D","created_at":"2026-06-30T02:16:53.654539+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":4,"internal_anchor_count":4,"sample":[{"citing_arxiv_id":"2601.10348","citing_title":"Training-Trajectory-Aware Token Selection","ref_index":23,"is_internal_anchor":true},{"citing_arxiv_id":"2601.13992","citing_title":"\"The Whole Is Greater Than the Sum of Its Parts\": A Compatibility-Aware Multi-Teacher CoT Distillation Framework","ref_index":13,"is_internal_anchor":true},{"citing_arxiv_id":"2605.19433","citing_title":"Backtracking When It Strays: Mitigating Dual Exposure Biases in LLM Reasoning Distillation","ref_index":39,"is_internal_anchor":true},{"citing_arxiv_id":"2601.14249","citing_title":"Which Reasoning Trajectories Teach Students to Reason Better? A Simple Metric of Informative Alignment","ref_index":38,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/THZM232D6HNPCOROAYDBJJHYDC","json":"https://pith.science/pith/THZM232D6HNPCOROAYDBJJHYDC.json","graph_json":"https://pith.science/api/pith-number/THZM232D6HNPCOROAYDBJJHYDC/graph.json","events_json":"https://pith.science/api/pith-number/THZM232D6HNPCOROAYDBJJHYDC/events.json","paper":"https://pith.science/paper/THZM232D"},"agent_actions":{"view_html":"https://pith.science/pith/THZM232D6HNPCOROAYDBJJHYDC","download_json":"https://pith.science/pith/THZM232D6HNPCOROAYDBJJHYDC.json","view_paper":"https://pith.science/paper/THZM232D","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2508.09883&json=true","fetch_graph":"https://pith.science/api/pith-number/THZM232D6HNPCOROAYDBJJHYDC/graph.json","fetch_events":"https://pith.science/api/pith-number/THZM232D6HNPCOROAYDBJJHYDC/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/THZM232D6HNPCOROAYDBJJHYDC/action/timestamp_anchor","attest_storage":"https://pith.science/pith/THZM232D6HNPCOROAYDBJJHYDC/action/storage_attestation","attest_author":"https://pith.science/pith/THZM232D6HNPCOROAYDBJJHYDC/action/author_attestation","sign_citation":"https://pith.science/pith/THZM232D6HNPCOROAYDBJJHYDC/action/citation_signature","submit_replication":"https://pith.science/pith/THZM232D6HNPCOROAYDBJJHYDC/action/replication_record"}},"created_at":"2026-06-30T02:16:53.654539+00:00","updated_at":"2026-06-30T02:16:53.654539+00:00"}