{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:AKZMNMFYBI6L6LXPZZFOGVXLS5","short_pith_number":"pith:AKZMNMFY","schema_version":"1.0","canonical_sha256":"02b2c6b0b80a3cbf2eefce4ae356eb975e9ff609d40075f06adf3e5c682b4aba","source":{"kind":"arxiv","id":"2505.24480","version":1},"attestation_state":"computed","paper":{"title":"Towards Effective Code-Integrated Reasoning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"Beichen Zhang, Fei Bai, Ji-Rong Wen, Lei Fang, Wayne Xin Zhao, Yingqian Min, Zheng Liu, Zhipeng Chen, Zhongyuan Wang","submitted_at":"2025-05-30T11:30:18Z","abstract_excerpt":"In this paper, we investigate code-integrated reasoning, where models generate code when necessary and integrate feedback by executing it through a code interpreter. To acquire this capability, models must learn when and how to use external code tools effectively, which is supported by tool-augmented reinforcement learning (RL) through interactive learning. Despite its benefits, tool-augmented RL can still suffer from potential instability in the learning dynamics. In light of this challenge, we present a systematic approach to improving the training effectiveness and stability of tool-augment"},"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":"2505.24480","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2025-05-30T11:30:18Z","cross_cats_sorted":["cs.AI","cs.LG"],"title_canon_sha256":"184121070caa72a462732f9f13ca946dcfa60d60e8820c7b7b569f34907e8ef4","abstract_canon_sha256":"8404f8fe96a6d16fd9389ea7f27628a80a84c7337d0bcb61947e6f72262d841e"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:12:50.764654Z","signature_b64":"0JtWTuwLYzQ27IkGXhcAlSusiLceI+USlumwdP8aAfarJiaQ/GmXX4tH5YY4xKfxPGho6rFTxk5Vy7/RgIl2Bw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"02b2c6b0b80a3cbf2eefce4ae356eb975e9ff609d40075f06adf3e5c682b4aba","last_reissued_at":"2026-07-05T11:12:50.764128Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:12:50.764128Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Towards Effective Code-Integrated Reasoning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"Beichen Zhang, Fei Bai, Ji-Rong Wen, Lei Fang, Wayne Xin Zhao, Yingqian Min, Zheng Liu, Zhipeng Chen, Zhongyuan Wang","submitted_at":"2025-05-30T11:30:18Z","abstract_excerpt":"In this paper, we investigate code-integrated reasoning, where models generate code when necessary and integrate feedback by executing it through a code interpreter. To acquire this capability, models must learn when and how to use external code tools effectively, which is supported by tool-augmented reinforcement learning (RL) through interactive learning. Despite its benefits, tool-augmented RL can still suffer from potential instability in the learning dynamics. In light of this challenge, we present a systematic approach to improving the training effectiveness and stability of tool-augment"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2505.24480","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/2505.24480/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":"2505.24480","created_at":"2026-07-05T11:12:50.764191+00:00"},{"alias_kind":"arxiv_version","alias_value":"2505.24480v1","created_at":"2026-07-05T11:12:50.764191+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2505.24480","created_at":"2026-07-05T11:12:50.764191+00:00"},{"alias_kind":"pith_short_12","alias_value":"AKZMNMFYBI6L","created_at":"2026-07-05T11:12:50.764191+00:00"},{"alias_kind":"pith_short_16","alias_value":"AKZMNMFYBI6L6LXP","created_at":"2026-07-05T11:12:50.764191+00:00"},{"alias_kind":"pith_short_8","alias_value":"AKZMNMFY","created_at":"2026-07-05T11:12:50.764191+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":4,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.19236","citing_title":"STARE: Surprisal-Guided Token-Level Advantage Reweighting for Policy Entropy Stability","ref_index":3,"is_internal_anchor":false},{"citing_arxiv_id":"2606.03762","citing_title":"Tool-Aware Optimization with Entropy Guidance for Efficient Agentic Reinforcement Learning","ref_index":12,"is_internal_anchor":false},{"citing_arxiv_id":"2605.09931","citing_title":"PruneTIR: Inference-Time Tool Call Pruning for Effective yet Efficient Tool-Integrated Reasoning","ref_index":38,"is_internal_anchor":false},{"citing_arxiv_id":"2604.08281","citing_title":"When to Trust Tools? Adaptive Tool Trust Calibration For Tool-Integrated Math Reasoning","ref_index":1,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/AKZMNMFYBI6L6LXPZZFOGVXLS5","json":"https://pith.science/pith/AKZMNMFYBI6L6LXPZZFOGVXLS5.json","graph_json":"https://pith.science/api/pith-number/AKZMNMFYBI6L6LXPZZFOGVXLS5/graph.json","events_json":"https://pith.science/api/pith-number/AKZMNMFYBI6L6LXPZZFOGVXLS5/events.json","paper":"https://pith.science/paper/AKZMNMFY"},"agent_actions":{"view_html":"https://pith.science/pith/AKZMNMFYBI6L6LXPZZFOGVXLS5","download_json":"https://pith.science/pith/AKZMNMFYBI6L6LXPZZFOGVXLS5.json","view_paper":"https://pith.science/paper/AKZMNMFY","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2505.24480&json=true","fetch_graph":"https://pith.science/api/pith-number/AKZMNMFYBI6L6LXPZZFOGVXLS5/graph.json","fetch_events":"https://pith.science/api/pith-number/AKZMNMFYBI6L6LXPZZFOGVXLS5/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/AKZMNMFYBI6L6LXPZZFOGVXLS5/action/timestamp_anchor","attest_storage":"https://pith.science/pith/AKZMNMFYBI6L6LXPZZFOGVXLS5/action/storage_attestation","attest_author":"https://pith.science/pith/AKZMNMFYBI6L6LXPZZFOGVXLS5/action/author_attestation","sign_citation":"https://pith.science/pith/AKZMNMFYBI6L6LXPZZFOGVXLS5/action/citation_signature","submit_replication":"https://pith.science/pith/AKZMNMFYBI6L6LXPZZFOGVXLS5/action/replication_record"}},"created_at":"2026-07-05T11:12:50.764191+00:00","updated_at":"2026-07-05T11:12:50.764191+00:00"}