{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:DBGUGFH3GZGVTGYIR7Q4SUVRPL","short_pith_number":"pith:DBGUGFH3","schema_version":"1.0","canonical_sha256":"184d4314fb364d599b088fe1c952b17af825c317d48b77593e681764fb77f476","source":{"kind":"arxiv","id":"2305.08844","version":2},"attestation_state":"computed","paper":{"title":"RL4F: Generating Natural Language Feedback with Reinforcement Learning for Repairing Model Outputs","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Afra Feyza Aky\\\"urek, Aman Madaan, Ashwin Kalyan, Derry Wijaya, Ekin Aky\\\"urek, Niket Tandon, Peter Clark","submitted_at":"2023-05-15T17:57:16Z","abstract_excerpt":"Despite their unprecedented success, even the largest language models make mistakes. Similar to how humans learn and improve using feedback, previous work proposed providing language models with natural language feedback to guide them in repairing their outputs. Because human-generated critiques are expensive to obtain, researchers have devised learned critique generators in lieu of human critics while assuming one can train downstream models to utilize generated feedback. However, this approach does not apply to black-box or limited access models such as ChatGPT, as they cannot be fine-tuned."},"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":"2305.08844","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2023-05-15T17:57:16Z","cross_cats_sorted":[],"title_canon_sha256":"e370ee5db532ed525f923a86565b6b6539e2e8c295fb7cee76d6d62c8b76d19c","abstract_canon_sha256":"88ce68ba8b4ef384516b247196a7eb97a79985b8790d69473890741cf232ba3b"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T06:30:07.037282Z","signature_b64":"fgFahp9RWpj7YN3uG/0EN+YcdVI2efGNFHYd+DtnOD+kE2u1DTmzms+kHHhs5e+q3rtnOcvngDDj3G58SvDVDQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"184d4314fb364d599b088fe1c952b17af825c317d48b77593e681764fb77f476","last_reissued_at":"2026-07-05T06:30:07.036876Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T06:30:07.036876Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"RL4F: Generating Natural Language Feedback with Reinforcement Learning for Repairing Model Outputs","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Afra Feyza Aky\\\"urek, Aman Madaan, Ashwin Kalyan, Derry Wijaya, Ekin Aky\\\"urek, Niket Tandon, Peter Clark","submitted_at":"2023-05-15T17:57:16Z","abstract_excerpt":"Despite their unprecedented success, even the largest language models make mistakes. Similar to how humans learn and improve using feedback, previous work proposed providing language models with natural language feedback to guide them in repairing their outputs. Because human-generated critiques are expensive to obtain, researchers have devised learned critique generators in lieu of human critics while assuming one can train downstream models to utilize generated feedback. However, this approach does not apply to black-box or limited access models such as ChatGPT, as they cannot be fine-tuned."},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2305.08844","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/2305.08844/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":"2305.08844","created_at":"2026-07-05T06:30:07.036936+00:00"},{"alias_kind":"arxiv_version","alias_value":"2305.08844v2","created_at":"2026-07-05T06:30:07.036936+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2305.08844","created_at":"2026-07-05T06:30:07.036936+00:00"},{"alias_kind":"pith_short_12","alias_value":"DBGUGFH3GZGV","created_at":"2026-07-05T06:30:07.036936+00:00"},{"alias_kind":"pith_short_16","alias_value":"DBGUGFH3GZGVTGYI","created_at":"2026-07-05T06:30:07.036936+00:00"},{"alias_kind":"pith_short_8","alias_value":"DBGUGFH3","created_at":"2026-07-05T06:30:07.036936+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.08068","citing_title":"DICE: Entropy-Regularized Equilibrium Selection for Stable Multi-Agent LLM Coordination","ref_index":99,"is_internal_anchor":false},{"citing_arxiv_id":"2409.12917","citing_title":"Training Language Models to Self-Correct via Reinforcement Learning","ref_index":2,"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":34,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/DBGUGFH3GZGVTGYIR7Q4SUVRPL","json":"https://pith.science/pith/DBGUGFH3GZGVTGYIR7Q4SUVRPL.json","graph_json":"https://pith.science/api/pith-number/DBGUGFH3GZGVTGYIR7Q4SUVRPL/graph.json","events_json":"https://pith.science/api/pith-number/DBGUGFH3GZGVTGYIR7Q4SUVRPL/events.json","paper":"https://pith.science/paper/DBGUGFH3"},"agent_actions":{"view_html":"https://pith.science/pith/DBGUGFH3GZGVTGYIR7Q4SUVRPL","download_json":"https://pith.science/pith/DBGUGFH3GZGVTGYIR7Q4SUVRPL.json","view_paper":"https://pith.science/paper/DBGUGFH3","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2305.08844&json=true","fetch_graph":"https://pith.science/api/pith-number/DBGUGFH3GZGVTGYIR7Q4SUVRPL/graph.json","fetch_events":"https://pith.science/api/pith-number/DBGUGFH3GZGVTGYIR7Q4SUVRPL/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/DBGUGFH3GZGVTGYIR7Q4SUVRPL/action/timestamp_anchor","attest_storage":"https://pith.science/pith/DBGUGFH3GZGVTGYIR7Q4SUVRPL/action/storage_attestation","attest_author":"https://pith.science/pith/DBGUGFH3GZGVTGYIR7Q4SUVRPL/action/author_attestation","sign_citation":"https://pith.science/pith/DBGUGFH3GZGVTGYIR7Q4SUVRPL/action/citation_signature","submit_replication":"https://pith.science/pith/DBGUGFH3GZGVTGYIR7Q4SUVRPL/action/replication_record"}},"created_at":"2026-07-05T06:30:07.036936+00:00","updated_at":"2026-07-05T06:30:07.036936+00:00"}