{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:LXCR2BOCI3ATVZTPUCRSAR2BX7","short_pith_number":"pith:LXCR2BOC","schema_version":"1.0","canonical_sha256":"5dc51d05c246c13ae66fa0a3204741bfd88ec5a865d5f1581372fbdb7741962b","source":{"kind":"arxiv","id":"2304.01904","version":2},"attestation_state":"computed","paper":{"title":"REFINER: Reasoning Feedback on Intermediate Representations","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Antoine Bosselut, Beatriz Borges, Boi Faltings, Debjit Paul, Maxime Peyrard, Mete Ismayilzada, Robert West","submitted_at":"2023-04-04T15:57:28Z","abstract_excerpt":"Language models (LMs) have recently shown remarkable performance on reasoning tasks by explicitly generating intermediate inferences, e.g., chain-of-thought prompting. However, these intermediate inference steps may be inappropriate deductions from the initial context and lead to incorrect final predictions. Here we introduce REFINER, a framework for finetuning LMs to explicitly generate intermediate reasoning steps while interacting with a critic model that provides automated feedback on the reasoning. Specifically, the critic provides structured feedback that the reasoning LM uses to iterati"},"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":"2304.01904","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2023-04-04T15:57:28Z","cross_cats_sorted":[],"title_canon_sha256":"145e8a0c1da93162c61ed4d48aaa45e0e0fee091b19b9ecbf9e3d2d77fe04679","abstract_canon_sha256":"944e3f19825e32d055d679b19d34fd023b56b1930a419d658f293a11b1c5b10f"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T07:40:52.606063Z","signature_b64":"kfC7Ha/mZMjmOSDPF8eekc7C7xmULxGrtxFjxLhXUbDvWw9hHxGGFtWmgqA7mPLFdJ/B2KiF/2D71JGgHV+SAA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"5dc51d05c246c13ae66fa0a3204741bfd88ec5a865d5f1581372fbdb7741962b","last_reissued_at":"2026-07-05T07:40:52.605653Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T07:40:52.605653Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"REFINER: Reasoning Feedback on Intermediate Representations","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Antoine Bosselut, Beatriz Borges, Boi Faltings, Debjit Paul, Maxime Peyrard, Mete Ismayilzada, Robert West","submitted_at":"2023-04-04T15:57:28Z","abstract_excerpt":"Language models (LMs) have recently shown remarkable performance on reasoning tasks by explicitly generating intermediate inferences, e.g., chain-of-thought prompting. However, these intermediate inference steps may be inappropriate deductions from the initial context and lead to incorrect final predictions. Here we introduce REFINER, a framework for finetuning LMs to explicitly generate intermediate reasoning steps while interacting with a critic model that provides automated feedback on the reasoning. Specifically, the critic provides structured feedback that the reasoning LM uses to iterati"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2304.01904","kind":"arxiv","version":2},"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/2304.01904/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":"2304.01904","created_at":"2026-07-05T07:40:52.605716+00:00"},{"alias_kind":"arxiv_version","alias_value":"2304.01904v2","created_at":"2026-07-05T07:40:52.605716+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2304.01904","created_at":"2026-07-05T07:40:52.605716+00:00"},{"alias_kind":"pith_short_12","alias_value":"LXCR2BOCI3AT","created_at":"2026-07-05T07:40:52.605716+00:00"},{"alias_kind":"pith_short_16","alias_value":"LXCR2BOCI3ATVZTP","created_at":"2026-07-05T07:40:52.605716+00:00"},{"alias_kind":"pith_short_8","alias_value":"LXCR2BOC","created_at":"2026-07-05T07:40:52.605716+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":14,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.17628","citing_title":"OPD-Evolver: Cultivating Holistic Agent Evolver via On-Policy Distillation","ref_index":117,"is_internal_anchor":false},{"citing_arxiv_id":"2606.08068","citing_title":"DICE: Entropy-Regularized Equilibrium Selection for Stable Multi-Agent LLM Coordination","ref_index":102,"is_internal_anchor":false},{"citing_arxiv_id":"2606.20728","citing_title":"VTOS: Learning to Orchestrate Vision Tools by Co-Searching Solutions and Observers","ref_index":15,"is_internal_anchor":false},{"citing_arxiv_id":"2508.15815","citing_title":"User-Assistant Bias in LLMs","ref_index":14,"is_internal_anchor":false},{"citing_arxiv_id":"2409.12917","citing_title":"Training Language Models to Self-Correct via Reinforcement Learning","ref_index":22,"is_internal_anchor":false},{"citing_arxiv_id":"2305.14992","citing_title":"Reasoning with Language Model is Planning with World Model","ref_index":95,"is_internal_anchor":false},{"citing_arxiv_id":"2602.02350","citing_title":"Context Learning for Multi-Agent Discussion","ref_index":15,"is_internal_anchor":false},{"citing_arxiv_id":"2310.11511","citing_title":"Self-RAG: Learning to Retrieve, Generate, and Critique through Self-Reflection","ref_index":37,"is_internal_anchor":false},{"citing_arxiv_id":"2310.01798","citing_title":"Large Language Models Cannot Self-Correct Reasoning Yet","ref_index":15,"is_internal_anchor":false},{"citing_arxiv_id":"2412.05579","citing_title":"LLMs-as-Judges: A Comprehensive Survey on LLM-based Evaluation Methods","ref_index":182,"is_internal_anchor":false},{"citing_arxiv_id":"2604.21268","citing_title":"Measure Twice, Click Once: Co-evolving Proposer and Visual Critic via Reinforcement Learning for GUI Grounding","ref_index":20,"is_internal_anchor":false},{"citing_arxiv_id":"2604.06066","citing_title":"From Hallucination to Structure Snowballing: The Alignment Tax of Constrained Decoding in LLM Reflection","ref_index":6,"is_internal_anchor":false},{"citing_arxiv_id":"2303.11366","citing_title":"Reflexion: Language Agents with Verbal Reinforcement Learning","ref_index":20,"is_internal_anchor":false},{"citing_arxiv_id":"2604.21057","citing_title":"TRACES: Tagging Reasoning Steps for Adaptive Cost-Efficient Early-Stopping","ref_index":4,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/LXCR2BOCI3ATVZTPUCRSAR2BX7","json":"https://pith.science/pith/LXCR2BOCI3ATVZTPUCRSAR2BX7.json","graph_json":"https://pith.science/api/pith-number/LXCR2BOCI3ATVZTPUCRSAR2BX7/graph.json","events_json":"https://pith.science/api/pith-number/LXCR2BOCI3ATVZTPUCRSAR2BX7/events.json","paper":"https://pith.science/paper/LXCR2BOC"},"agent_actions":{"view_html":"https://pith.science/pith/LXCR2BOCI3ATVZTPUCRSAR2BX7","download_json":"https://pith.science/pith/LXCR2BOCI3ATVZTPUCRSAR2BX7.json","view_paper":"https://pith.science/paper/LXCR2BOC","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2304.01904&json=true","fetch_graph":"https://pith.science/api/pith-number/LXCR2BOCI3ATVZTPUCRSAR2BX7/graph.json","fetch_events":"https://pith.science/api/pith-number/LXCR2BOCI3ATVZTPUCRSAR2BX7/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/LXCR2BOCI3ATVZTPUCRSAR2BX7/action/timestamp_anchor","attest_storage":"https://pith.science/pith/LXCR2BOCI3ATVZTPUCRSAR2BX7/action/storage_attestation","attest_author":"https://pith.science/pith/LXCR2BOCI3ATVZTPUCRSAR2BX7/action/author_attestation","sign_citation":"https://pith.science/pith/LXCR2BOCI3ATVZTPUCRSAR2BX7/action/citation_signature","submit_replication":"https://pith.science/pith/LXCR2BOCI3ATVZTPUCRSAR2BX7/action/replication_record"}},"created_at":"2026-07-05T07:40:52.605716+00:00","updated_at":"2026-07-05T07:40:52.605716+00:00"}