{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:3WE7LPRTJJF7PEZ2S25A2IAP3W","short_pith_number":"pith:3WE7LPRT","schema_version":"1.0","canonical_sha256":"dd89f5be334a4bf7933a96ba0d200fdda07c40408dee4eb6a83fba23befb2395","source":{"kind":"arxiv","id":"2308.01825","version":2},"attestation_state":"computed","paper":{"title":"Scaling Relationship on Learning Mathematical Reasoning with Large Language Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"Pre-training loss predicts LLM mathematical reasoning performance better than parameter count, and rejection sampling fine-tuning lifts LLaMA-7B to 49.3 percent accuracy on GSM8K.","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Chang Zhou, Chengpeng Li, Chuanqi Tan, Guanting Dong, Hongyi Yuan, Jingren Zhou, Keming Lu, Zheng Yuan","submitted_at":"2023-08-03T15:34:01Z","abstract_excerpt":"Mathematical reasoning is a challenging task for large language models (LLMs), while the scaling relationship of it with respect to LLM capacity is under-explored. In this paper, we investigate how the pre-training loss, supervised data amount, and augmented data amount influence the reasoning performances of a supervised LLM. We find that pre-training loss is a better indicator of the model's performance than the model's parameter count. We apply supervised fine-tuning (SFT) with different amounts of supervised data and empirically find a log-linear relation between data amount and model perf"},"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":true,"formal_links_present":true},"canonical_record":{"source":{"id":"2308.01825","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2023-08-03T15:34:01Z","cross_cats_sorted":[],"title_canon_sha256":"c762f07c2dee1db9acd2261bf247e9a3ef1c3108732525103ccfca0718b6a249","abstract_canon_sha256":"1b9def5d64481b95ede3a3744d495f79937b5bf1849f35e0103828ced38257f7"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-17T23:39:19.735122Z","signature_b64":"ZryLUoUruXgyHhU8xmXanqFoQB635o5hASz0xcMjzhyPoS1++C0XV/cDs3iPWQB7YXzt6Zk20d1wLggSR5tRAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"dd89f5be334a4bf7933a96ba0d200fdda07c40408dee4eb6a83fba23befb2395","last_reissued_at":"2026-05-17T23:39:19.734436Z","signature_status":"signed_v1","first_computed_at":"2026-05-17T23:39:19.734436Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Scaling Relationship on Learning Mathematical Reasoning with Large Language Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"Pre-training loss predicts LLM mathematical reasoning performance better than parameter count, and rejection sampling fine-tuning lifts LLaMA-7B to 49.3 percent accuracy on GSM8K.","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Chang Zhou, Chengpeng Li, Chuanqi Tan, Guanting Dong, Hongyi Yuan, Jingren Zhou, Keming Lu, Zheng Yuan","submitted_at":"2023-08-03T15:34:01Z","abstract_excerpt":"Mathematical reasoning is a challenging task for large language models (LLMs), while the scaling relationship of it with respect to LLM capacity is under-explored. In this paper, we investigate how the pre-training loss, supervised data amount, and augmented data amount influence the reasoning performances of a supervised LLM. We find that pre-training loss is a better indicator of the model's performance than the model's parameter count. We apply supervised fine-tuning (SFT) with different amounts of supervised data and empirically find a log-linear relation between data amount and model perf"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"We find that pre-training loss is a better indicator of the model's performance than the model's parameter count. ... we combine rejection samples from multiple models which push LLaMA-7B to an accuracy of 49.3% on GSM8K which outperforms the supervised fine-tuning (SFT) accuracy of 35.9% significantly.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That model-generated reasoning paths can be reliably verified as correct by the same or similar models without introducing systematic errors or false positives in the rejection filter.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"Pre-training loss predicts LLM math reasoning better than parameter count; rejection sampling fine-tuning with diverse paths raises LLaMA-7B accuracy on GSM8K from 35.9% with SFT to 49.3%.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Pre-training loss predicts LLM mathematical reasoning performance better than parameter count, and rejection sampling fine-tuning lifts LLaMA-7B to 49.3 percent accuracy on GSM8K.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"b9885aca34cf9dbad23e33ae094b469737b335d61fd858b088cc95993e59f052"},"source":{"id":"2308.01825","kind":"arxiv","version":2},"verdict":{"id":"24aeb9b5-30b3-4909-ae7e-b9f017abcce0","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-15T00:16:47.344407Z","strongest_claim":"We find that pre-training loss is a better indicator of the model's performance than the model's parameter count. ... we combine rejection samples from multiple models which push LLaMA-7B to an accuracy of 49.3% on GSM8K which outperforms the supervised fine-tuning (SFT) accuracy of 35.9% significantly.","one_line_summary":"Pre-training loss predicts LLM math reasoning better than parameter count; rejection sampling fine-tuning with diverse paths raises LLaMA-7B accuracy on GSM8K from 35.9% with SFT to 49.3%.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That model-generated reasoning paths can be reliably verified as correct by the same or similar models without introducing systematic errors or false positives in the rejection filter.","pith_extraction_headline":"Pre-training loss predicts LLM mathematical reasoning performance better than parameter count, and rejection sampling fine-tuning lifts LLaMA-7B to 49.3 percent accuracy on GSM8K."},"references":{"count":93,"sample":[{"doi":"","year":null,"title":"Emergent Abilities of Large Language Models , author=. Trans. Mach. Learn. Res. , year=","work_id":"e7481617-8603-477e-bfea-e36897ba0381","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":null,"title":"Finetuned Language Models Are Zero-Shot Learners , author=. ArXiv , year=","work_id":"ca4b85c3-127f-42a0-83c7-263974e3a2fd","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":null,"title":"Chain of Thought Prompting Elicits Reasoning in Large Language Models , author=. ArXiv , year=","work_id":"75eacb63-2829-4757-8923-44fbe8ccdd56","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":null,"title":"The Eleventh International Conference on Learning Representations , year=","work_id":"c6ff7dc0-3e26-4410-9185-4ed25c298501","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2023,"title":"Scaling Data-Constrained Language Models , author=. 2023 , eprint=","work_id":"3d681b07-4c9c-441c-b5f3-8efb06f3656a","ref_index":6,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":93,"snapshot_sha256":"3d1e11b8e1561056585bf6cb1192ba3b63516f553cc7ba8c0cdda9e5ba2e0afc","internal_anchors":10},"formal_canon":{"evidence_count":2,"snapshot_sha256":"3953eebc194b1a96dee1e12ae9ba23642cdcb93447eb1a14fa22dd78f9d447e4"},"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":"2308.01825","created_at":"2026-05-17T23:39:19.734542+00:00"},{"alias_kind":"arxiv_version","alias_value":"2308.01825v2","created_at":"2026-05-17T23:39:19.734542+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2308.01825","created_at":"2026-05-17T23:39:19.734542+00:00"},{"alias_kind":"pith_short_12","alias_value":"3WE7LPRTJJF7","created_at":"2026-05-18T12:33:33.725879+00:00"},{"alias_kind":"pith_short_16","alias_value":"3WE7LPRTJJF7PEZ2","created_at":"2026-05-18T12:33:33.725879+00:00"},{"alias_kind":"pith_short_8","alias_value":"3WE7LPRT","created_at":"2026-05-18T12:33:33.725879+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":42,"internal_anchor_count":42,"sample":[{"citing_arxiv_id":"2402.03300","citing_title":"DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models","ref_index":57,"is_internal_anchor":true},{"citing_arxiv_id":"2412.04468","citing_title":"NVILA: Efficient Frontier Visual Language Models","ref_index":138,"is_internal_anchor":true},{"citing_arxiv_id":"2412.15251","citing_title":"IPS: In-Prompt Process Supervision for Short Video Content Moderation","ref_index":2,"is_internal_anchor":true},{"citing_arxiv_id":"2412.15115","citing_title":"Qwen2.5 Technical Report","ref_index":43,"is_internal_anchor":true},{"citing_arxiv_id":"2504.02181","citing_title":"A Survey of Scaling in Large Language Model Reasoning","ref_index":248,"is_internal_anchor":true},{"citing_arxiv_id":"2510.07962","citing_title":"LightReasoner: Can Small Language Models Teach Large Language Models Reasoning?","ref_index":20,"is_internal_anchor":true},{"citing_arxiv_id":"2605.22389","citing_title":"Unified Data Selection for LLM Reasoning","ref_index":35,"is_internal_anchor":true},{"citing_arxiv_id":"2605.22389","citing_title":"Unified Data Selection for LLM Reasoning","ref_index":33,"is_internal_anchor":true},{"citing_arxiv_id":"2410.08146","citing_title":"Rewarding Progress: Scaling Automated Process Verifiers for LLM Reasoning","ref_index":24,"is_internal_anchor":true},{"citing_arxiv_id":"2309.17452","citing_title":"ToRA: A Tool-Integrated Reasoning Agent for Mathematical Problem Solving","ref_index":47,"is_internal_anchor":true},{"citing_arxiv_id":"2502.10248","citing_title":"Step-Video-T2V Technical Report: The Practice, Challenges, and Future of Video Foundation Model","ref_index":121,"is_internal_anchor":true},{"citing_arxiv_id":"2502.03373","citing_title":"Demystifying Long Chain-of-Thought Reasoning in LLMs","ref_index":5,"is_internal_anchor":true},{"citing_arxiv_id":"2406.18629","citing_title":"Step-DPO: Step-wise Preference Optimization for Long-chain Reasoning of LLMs","ref_index":33,"is_internal_anchor":true},{"citing_arxiv_id":"2505.15134","citing_title":"The Unreasonable Effectiveness of Entropy Minimization in LLM Reasoning","ref_index":95,"is_internal_anchor":true},{"citing_arxiv_id":"2509.21543","citing_title":"Self-CriTeach: LLM Self-Teaching and Self-Critiquing for Improving Robotic Planning via Automated Domain Generation","ref_index":21,"is_internal_anchor":true},{"citing_arxiv_id":"2509.23330","citing_title":"Structured In-context Environment Scaling for Large Language Model Reasoning","ref_index":28,"is_internal_anchor":true},{"citing_arxiv_id":"2412.09413","citing_title":"Imitate, Explore, and Self-Improve: A Reproduction Report on Slow-thinking Reasoning Systems","ref_index":26,"is_internal_anchor":true},{"citing_arxiv_id":"2309.05653","citing_title":"MAmmoTH: Building Math Generalist Models through Hybrid Instruction Tuning","ref_index":68,"is_internal_anchor":true},{"citing_arxiv_id":"2507.19849","citing_title":"Agentic Reinforced Policy Optimization","ref_index":5,"is_internal_anchor":true},{"citing_arxiv_id":"2505.18719","citing_title":"VLA-RL: Towards Masterful and General Robotic Manipulation with Scalable Reinforcement Learning","ref_index":83,"is_internal_anchor":true},{"citing_arxiv_id":"2603.26499","citing_title":"AIRA_2: Overcoming Bottlenecks in AI Research Agents","ref_index":23,"is_internal_anchor":true},{"citing_arxiv_id":"2312.08935","citing_title":"Math-Shepherd: Verify and Reinforce LLMs Step-by-step without Human Annotations","ref_index":92,"is_internal_anchor":true},{"citing_arxiv_id":"2604.00931","citing_title":"PsychAgent: An Experience-Driven Lifelong Learning Agent for Self-Evolving Psychological Counselor","ref_index":35,"is_internal_anchor":true},{"citing_arxiv_id":"2501.05366","citing_title":"Search-o1: Agentic Search-Enhanced Large Reasoning Models","ref_index":74,"is_internal_anchor":true},{"citing_arxiv_id":"2309.12284","citing_title":"MetaMath: Bootstrap Your Own Mathematical Questions for Large Language Models","ref_index":79,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":2,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/3WE7LPRTJJF7PEZ2S25A2IAP3W","json":"https://pith.science/pith/3WE7LPRTJJF7PEZ2S25A2IAP3W.json","graph_json":"https://pith.science/api/pith-number/3WE7LPRTJJF7PEZ2S25A2IAP3W/graph.json","events_json":"https://pith.science/api/pith-number/3WE7LPRTJJF7PEZ2S25A2IAP3W/events.json","paper":"https://pith.science/paper/3WE7LPRT"},"agent_actions":{"view_html":"https://pith.science/pith/3WE7LPRTJJF7PEZ2S25A2IAP3W","download_json":"https://pith.science/pith/3WE7LPRTJJF7PEZ2S25A2IAP3W.json","view_paper":"https://pith.science/paper/3WE7LPRT","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2308.01825&json=true","fetch_graph":"https://pith.science/api/pith-number/3WE7LPRTJJF7PEZ2S25A2IAP3W/graph.json","fetch_events":"https://pith.science/api/pith-number/3WE7LPRTJJF7PEZ2S25A2IAP3W/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/3WE7LPRTJJF7PEZ2S25A2IAP3W/action/timestamp_anchor","attest_storage":"https://pith.science/pith/3WE7LPRTJJF7PEZ2S25A2IAP3W/action/storage_attestation","attest_author":"https://pith.science/pith/3WE7LPRTJJF7PEZ2S25A2IAP3W/action/author_attestation","sign_citation":"https://pith.science/pith/3WE7LPRTJJF7PEZ2S25A2IAP3W/action/citation_signature","submit_replication":"https://pith.science/pith/3WE7LPRTJJF7PEZ2S25A2IAP3W/action/replication_record"}},"created_at":"2026-05-17T23:39:19.734542+00:00","updated_at":"2026-05-17T23:39:19.734542+00:00"}