{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:74ZB5UNA7IGYDELU5HCGFQF57Z","short_pith_number":"pith:74ZB5UNA","schema_version":"1.0","canonical_sha256":"ff321ed1a0fa0d819174e9c462c0bdfe5e34252af2e62140df767275542e957f","source":{"kind":"arxiv","id":"2412.16871","version":1},"attestation_state":"computed","paper":{"title":"Teaching LLMs to Refine with Tools","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Dian Yu, Dong Yu, Haitao Mi, Jiahao Xu, Linfeng Song, Tian Liang, Yuheng Zhang, Zhaopeng Tu","submitted_at":"2024-12-22T05:43:50Z","abstract_excerpt":"Large language models (LLMs) can refine their responses based on feedback, enabling self-improvement through iterative training or test-time refinement. However, existing methods predominantly focus on refinement within the same reasoning format, which may lead to non-correcting behaviors. We propose CaP, a novel approach that uses external tools to refine chain-of-thought (CoT) responses generated by the same or other LLMs. CaP employs a two-stage training process: supervised fine-tuning followed by preference optimization with DPO variants. Our observations highlight the critical role of pre"},"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":"2412.16871","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-12-22T05:43:50Z","cross_cats_sorted":[],"title_canon_sha256":"ec258769838e10f7446c767f5b608bcd29b5a2d3cf886a804b656d3b8aa2b62c","abstract_canon_sha256":"6af52ed7799d1524c021b06ef479f26100a6445a8987146a601d2acd69ee04b5"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:53:13.983416Z","signature_b64":"trQ/hxyscUyO48/X2IB7jdulhDnlewZtYTOc3RTPI7VIXFZQawUktOVWrGtqQQnH9CNoCKEaMEoDAoZWia5OBw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"ff321ed1a0fa0d819174e9c462c0bdfe5e34252af2e62140df767275542e957f","last_reissued_at":"2026-07-05T09:53:13.982934Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:53:13.982934Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Teaching LLMs to Refine with Tools","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Dian Yu, Dong Yu, Haitao Mi, Jiahao Xu, Linfeng Song, Tian Liang, Yuheng Zhang, Zhaopeng Tu","submitted_at":"2024-12-22T05:43:50Z","abstract_excerpt":"Large language models (LLMs) can refine their responses based on feedback, enabling self-improvement through iterative training or test-time refinement. However, existing methods predominantly focus on refinement within the same reasoning format, which may lead to non-correcting behaviors. We propose CaP, a novel approach that uses external tools to refine chain-of-thought (CoT) responses generated by the same or other LLMs. CaP employs a two-stage training process: supervised fine-tuning followed by preference optimization with DPO variants. Our observations highlight the critical role of pre"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2412.16871","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/2412.16871/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":"2412.16871","created_at":"2026-07-05T09:53:13.982992+00:00"},{"alias_kind":"arxiv_version","alias_value":"2412.16871v1","created_at":"2026-07-05T09:53:13.982992+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2412.16871","created_at":"2026-07-05T09:53:13.982992+00:00"},{"alias_kind":"pith_short_12","alias_value":"74ZB5UNA7IGY","created_at":"2026-07-05T09:53:13.982992+00:00"},{"alias_kind":"pith_short_16","alias_value":"74ZB5UNA7IGYDELU","created_at":"2026-07-05T09:53:13.982992+00:00"},{"alias_kind":"pith_short_8","alias_value":"74ZB5UNA","created_at":"2026-07-05T09:53:13.982992+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/74ZB5UNA7IGYDELU5HCGFQF57Z","json":"https://pith.science/pith/74ZB5UNA7IGYDELU5HCGFQF57Z.json","graph_json":"https://pith.science/api/pith-number/74ZB5UNA7IGYDELU5HCGFQF57Z/graph.json","events_json":"https://pith.science/api/pith-number/74ZB5UNA7IGYDELU5HCGFQF57Z/events.json","paper":"https://pith.science/paper/74ZB5UNA"},"agent_actions":{"view_html":"https://pith.science/pith/74ZB5UNA7IGYDELU5HCGFQF57Z","download_json":"https://pith.science/pith/74ZB5UNA7IGYDELU5HCGFQF57Z.json","view_paper":"https://pith.science/paper/74ZB5UNA","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2412.16871&json=true","fetch_graph":"https://pith.science/api/pith-number/74ZB5UNA7IGYDELU5HCGFQF57Z/graph.json","fetch_events":"https://pith.science/api/pith-number/74ZB5UNA7IGYDELU5HCGFQF57Z/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/74ZB5UNA7IGYDELU5HCGFQF57Z/action/timestamp_anchor","attest_storage":"https://pith.science/pith/74ZB5UNA7IGYDELU5HCGFQF57Z/action/storage_attestation","attest_author":"https://pith.science/pith/74ZB5UNA7IGYDELU5HCGFQF57Z/action/author_attestation","sign_citation":"https://pith.science/pith/74ZB5UNA7IGYDELU5HCGFQF57Z/action/citation_signature","submit_replication":"https://pith.science/pith/74ZB5UNA7IGYDELU5HCGFQF57Z/action/replication_record"}},"created_at":"2026-07-05T09:53:13.982992+00:00","updated_at":"2026-07-05T09:53:13.982992+00:00"}