{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:ACBHUCGXZBSI24D74G5LGYIL57","short_pith_number":"pith:ACBHUCGX","schema_version":"1.0","canonical_sha256":"00827a08d7c8648d707fe1bab3610befcddd07620b916d5c7ae9ee40bcd0cd7c","source":{"kind":"arxiv","id":"2410.07745","version":4},"attestation_state":"computed","paper":{"title":"StepTool: Enhancing Multi-Step Tool Usage in LLMs via Step-Grained Reinforcement Learning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Chuhan Wu, Min Zhang, Shuai Wang, Weizhi Ma, Yuanqing Yu, Zhefan Wang, Zhiqiang Guo","submitted_at":"2024-10-10T09:23:26Z","abstract_excerpt":"Despite their powerful text generation capabilities, large language models (LLMs) still struggle to effectively utilize external tools to solve complex tasks, a challenge known as tool learning. Existing methods primarily rely on supervised fine-tuning, treating tool learning as a text generation problem while overlooking the decision-making complexities inherent in multi-step contexts. In this work, we propose modeling tool learning as a dynamic decision-making process and introduce StepTool, a novel step-grained reinforcement learning framework that enhances LLMs' capabilities in multi-step "},"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":"2410.07745","kind":"arxiv","version":4},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-10-10T09:23:26Z","cross_cats_sorted":[],"title_canon_sha256":"418ce688f778ba91d54dc7173c7c930a32c6eadd51ed3dcc021d32fa6016d092","abstract_canon_sha256":"2c3213bc4e9b557378029e992701f2b1422fa2b5c5f2a43f4d7562a5e48067c1"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:54:42.046129Z","signature_b64":"xp4fwKKGKzKkh4L3eTywKGMPYnvPbIh4YfOITG5OSrU6/3Mu2B21EKhIdENdQFCjhmn2Y3/01l73pSb8AczWCw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"00827a08d7c8648d707fe1bab3610befcddd07620b916d5c7ae9ee40bcd0cd7c","last_reissued_at":"2026-07-05T11:54:42.045633Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:54:42.045633Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"StepTool: Enhancing Multi-Step Tool Usage in LLMs via Step-Grained Reinforcement Learning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Chuhan Wu, Min Zhang, Shuai Wang, Weizhi Ma, Yuanqing Yu, Zhefan Wang, Zhiqiang Guo","submitted_at":"2024-10-10T09:23:26Z","abstract_excerpt":"Despite their powerful text generation capabilities, large language models (LLMs) still struggle to effectively utilize external tools to solve complex tasks, a challenge known as tool learning. Existing methods primarily rely on supervised fine-tuning, treating tool learning as a text generation problem while overlooking the decision-making complexities inherent in multi-step contexts. In this work, we propose modeling tool learning as a dynamic decision-making process and introduce StepTool, a novel step-grained reinforcement learning framework that enhances LLMs' capabilities in multi-step "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2410.07745","kind":"arxiv","version":4},"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/2410.07745/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":"2410.07745","created_at":"2026-07-05T11:54:42.045692+00:00"},{"alias_kind":"arxiv_version","alias_value":"2410.07745v4","created_at":"2026-07-05T11:54:42.045692+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2410.07745","created_at":"2026-07-05T11:54:42.045692+00:00"},{"alias_kind":"pith_short_12","alias_value":"ACBHUCGXZBSI","created_at":"2026-07-05T11:54:42.045692+00:00"},{"alias_kind":"pith_short_16","alias_value":"ACBHUCGXZBSI24D7","created_at":"2026-07-05T11:54:42.045692+00:00"},{"alias_kind":"pith_short_8","alias_value":"ACBHUCGX","created_at":"2026-07-05T11:54:42.045692+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":6,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2607.07435","citing_title":"RLVP: Penalize the Path, Reward the Outcome","ref_index":28,"is_internal_anchor":true},{"citing_arxiv_id":"2606.21228","citing_title":"Sakana Fugu Technical Report","ref_index":297,"is_internal_anchor":false},{"citing_arxiv_id":"2606.03892","citing_title":"Synthesize and Reward -- Reinforcement Learning for Multi-Step Tool Use in Live Environments","ref_index":24,"is_internal_anchor":false},{"citing_arxiv_id":"2605.27788","citing_title":"Knowing When to Ask: Segment-Level Credit Assignment for LLM Tool Use","ref_index":27,"is_internal_anchor":false},{"citing_arxiv_id":"2508.08791","citing_title":"Feedback-Driven Tool-Use Improvements in Large Language Models via Automated Build Environments","ref_index":57,"is_internal_anchor":false},{"citing_arxiv_id":"2504.13958","citing_title":"ToolRL: Reward is All Tool Learning Needs","ref_index":39,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/ACBHUCGXZBSI24D74G5LGYIL57","json":"https://pith.science/pith/ACBHUCGXZBSI24D74G5LGYIL57.json","graph_json":"https://pith.science/api/pith-number/ACBHUCGXZBSI24D74G5LGYIL57/graph.json","events_json":"https://pith.science/api/pith-number/ACBHUCGXZBSI24D74G5LGYIL57/events.json","paper":"https://pith.science/paper/ACBHUCGX"},"agent_actions":{"view_html":"https://pith.science/pith/ACBHUCGXZBSI24D74G5LGYIL57","download_json":"https://pith.science/pith/ACBHUCGXZBSI24D74G5LGYIL57.json","view_paper":"https://pith.science/paper/ACBHUCGX","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2410.07745&json=true","fetch_graph":"https://pith.science/api/pith-number/ACBHUCGXZBSI24D74G5LGYIL57/graph.json","fetch_events":"https://pith.science/api/pith-number/ACBHUCGXZBSI24D74G5LGYIL57/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/ACBHUCGXZBSI24D74G5LGYIL57/action/timestamp_anchor","attest_storage":"https://pith.science/pith/ACBHUCGXZBSI24D74G5LGYIL57/action/storage_attestation","attest_author":"https://pith.science/pith/ACBHUCGXZBSI24D74G5LGYIL57/action/author_attestation","sign_citation":"https://pith.science/pith/ACBHUCGXZBSI24D74G5LGYIL57/action/citation_signature","submit_replication":"https://pith.science/pith/ACBHUCGXZBSI24D74G5LGYIL57/action/replication_record"}},"created_at":"2026-07-05T11:54:42.045692+00:00","updated_at":"2026-07-05T11:54:42.045692+00:00"}