{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:QHDATL6PRPHOOQEJOJXIAGEP2R","short_pith_number":"pith:QHDATL6P","schema_version":"1.0","canonical_sha256":"81c609afcf8bcee74089726e80188fd45b420521cc68725ab2c7477629df50a1","source":{"kind":"arxiv","id":"2406.03816","version":3},"attestation_state":"computed","paper":{"title":"ReST-MCTS*: LLM Self-Training via Process Reward Guided Tree Search","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Dan Zhang, Jie Tang, Sining Zhoubian, Yisong Yue, Yuxiao Dong, Ziniu Hu","submitted_at":"2024-06-06T07:40:00Z","abstract_excerpt":"Recent methodologies in LLM self-training mostly rely on LLM generating responses and filtering those with correct output answers as training data. This approach often yields a low-quality fine-tuning training set (e.g., incorrect plans or intermediate reasoning). In this paper, we develop a reinforced self-training approach, called ReST-MCTS*, based on integrating process reward guidance with tree search MCTS* for collecting higher-quality reasoning traces as well as per-step value to train policy and reward models. ReST-MCTS* circumvents the per-step manual annotation typically used to train"},"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":"2406.03816","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-06-06T07:40:00Z","cross_cats_sorted":[],"title_canon_sha256":"e0193e56630e62ecefc5e26869b0d34823cf01b3f95d8167785825e3404eb6e0","abstract_canon_sha256":"09d139f3b9f099fa0afcab3c9365bb0c8f1a88174f9455665dfd43170a5e94f0"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:36:41.361910Z","signature_b64":"O0eabWi+CZi5Z+yZEH3a+TnpbWhRwZx+/VOg9MJ9z/FlboD6Om29ST/+9kvLdKdseqKUx+AiNmNTtnMEPFPcDg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"81c609afcf8bcee74089726e80188fd45b420521cc68725ab2c7477629df50a1","last_reissued_at":"2026-07-05T09:36:41.361420Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:36:41.361420Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"ReST-MCTS*: LLM Self-Training via Process Reward Guided Tree Search","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Dan Zhang, Jie Tang, Sining Zhoubian, Yisong Yue, Yuxiao Dong, Ziniu Hu","submitted_at":"2024-06-06T07:40:00Z","abstract_excerpt":"Recent methodologies in LLM self-training mostly rely on LLM generating responses and filtering those with correct output answers as training data. This approach often yields a low-quality fine-tuning training set (e.g., incorrect plans or intermediate reasoning). In this paper, we develop a reinforced self-training approach, called ReST-MCTS*, based on integrating process reward guidance with tree search MCTS* for collecting higher-quality reasoning traces as well as per-step value to train policy and reward models. ReST-MCTS* circumvents the per-step manual annotation typically used to train"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2406.03816","kind":"arxiv","version":3},"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/2406.03816/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":"2406.03816","created_at":"2026-07-05T09:36:41.361477+00:00"},{"alias_kind":"arxiv_version","alias_value":"2406.03816v3","created_at":"2026-07-05T09:36:41.361477+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2406.03816","created_at":"2026-07-05T09:36:41.361477+00:00"},{"alias_kind":"pith_short_12","alias_value":"QHDATL6PRPHO","created_at":"2026-07-05T09:36:41.361477+00:00"},{"alias_kind":"pith_short_16","alias_value":"QHDATL6PRPHOOQEJ","created_at":"2026-07-05T09:36:41.361477+00:00"},{"alias_kind":"pith_short_8","alias_value":"QHDATL6P","created_at":"2026-07-05T09:36:41.361477+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":12,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2607.07663","citing_title":"Recursive Self-Improvement in AI: From Bounded Self-Refinement to Autonomous Research Loops","ref_index":89,"is_internal_anchor":true},{"citing_arxiv_id":"2607.01223","citing_title":"Theoria: Rewrite-Acceptability Verification over Informal Reasoning States","ref_index":14,"is_internal_anchor":false},{"citing_arxiv_id":"2606.05464","citing_title":"Step-by-Step Optimization-like Reasoning in LLMs over Expanding Search Spaces","ref_index":41,"is_internal_anchor":false},{"citing_arxiv_id":"2606.00618","citing_title":"Efficient Test-time Inference for Generative Planning Models with OCL Search","ref_index":44,"is_internal_anchor":false},{"citing_arxiv_id":"2503.12937","citing_title":"R1-VL: Learning to Reason with Multimodal Large Language Models via Step-wise Group Relative Policy Optimization","ref_index":50,"is_internal_anchor":false},{"citing_arxiv_id":"2501.09686","citing_title":"Towards Large Reasoning Models: A Survey of Reinforced Reasoning with Large Language Models","ref_index":185,"is_internal_anchor":false},{"citing_arxiv_id":"2406.06592","citing_title":"Improve Mathematical Reasoning in Language Models by Automated Process Supervision","ref_index":26,"is_internal_anchor":false},{"citing_arxiv_id":"2502.17419","citing_title":"From System 1 to System 2: A Survey of Reasoning Large Language Models","ref_index":172,"is_internal_anchor":false},{"citing_arxiv_id":"2605.03625","citing_title":"Self-Improvement for Fast, High-Quality Plan Generation","ref_index":38,"is_internal_anchor":false},{"citing_arxiv_id":"2605.05262","citing_title":"Maximizing Rollout Informativeness under a Fixed Budget: A Submodular View of Tree Search for Tool-Use Agentic Reinforcement Learning","ref_index":25,"is_internal_anchor":false},{"citing_arxiv_id":"2604.10827","citing_title":"Your Model Diversity, Not Method, Determines Reasoning Strategy","ref_index":19,"is_internal_anchor":false},{"citing_arxiv_id":"2604.18327","citing_title":"PARM: Pipeline-Adapted Reward Model","ref_index":7,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/QHDATL6PRPHOOQEJOJXIAGEP2R","json":"https://pith.science/pith/QHDATL6PRPHOOQEJOJXIAGEP2R.json","graph_json":"https://pith.science/api/pith-number/QHDATL6PRPHOOQEJOJXIAGEP2R/graph.json","events_json":"https://pith.science/api/pith-number/QHDATL6PRPHOOQEJOJXIAGEP2R/events.json","paper":"https://pith.science/paper/QHDATL6P"},"agent_actions":{"view_html":"https://pith.science/pith/QHDATL6PRPHOOQEJOJXIAGEP2R","download_json":"https://pith.science/pith/QHDATL6PRPHOOQEJOJXIAGEP2R.json","view_paper":"https://pith.science/paper/QHDATL6P","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2406.03816&json=true","fetch_graph":"https://pith.science/api/pith-number/QHDATL6PRPHOOQEJOJXIAGEP2R/graph.json","fetch_events":"https://pith.science/api/pith-number/QHDATL6PRPHOOQEJOJXIAGEP2R/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/QHDATL6PRPHOOQEJOJXIAGEP2R/action/timestamp_anchor","attest_storage":"https://pith.science/pith/QHDATL6PRPHOOQEJOJXIAGEP2R/action/storage_attestation","attest_author":"https://pith.science/pith/QHDATL6PRPHOOQEJOJXIAGEP2R/action/author_attestation","sign_citation":"https://pith.science/pith/QHDATL6PRPHOOQEJOJXIAGEP2R/action/citation_signature","submit_replication":"https://pith.science/pith/QHDATL6PRPHOOQEJOJXIAGEP2R/action/replication_record"}},"created_at":"2026-07-05T09:36:41.361477+00:00","updated_at":"2026-07-05T09:36:41.361477+00:00"}