{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:PBQMKTSOADE6E2P2SJHYVCAC6W","short_pith_number":"pith:PBQMKTSO","schema_version":"1.0","canonical_sha256":"7860c54e4e00c9e269fa924f8a8802f589121963e30d080d10e948be641226c2","source":{"kind":"arxiv","id":"2402.10893","version":1},"attestation_state":"computed","paper":{"title":"RLVF: Learning from Verbal Feedback without Overgeneralization","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.LG","authors_text":"Alexander Khazatsky, Annie S Chen, Archit Sharma, Chelsea Finn, Eric Mitchell, Moritz Stephan, Sheryl Hsu","submitted_at":"2024-02-16T18:50:24Z","abstract_excerpt":"The diversity of contexts in which large language models (LLMs) are deployed requires the ability to modify or customize default model behaviors to incorporate nuanced requirements and preferences. A convenient interface to specify such model adjustments is high-level verbal feedback, such as \"Don't use emojis when drafting emails to my boss.\" However, while writing high-level feedback is far simpler than collecting annotations for reinforcement learning from human feedback (RLHF), we find that simply prompting a model with such feedback leads to overgeneralization of the feedback to contexts "},"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":"2402.10893","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-02-16T18:50:24Z","cross_cats_sorted":["cs.AI","cs.CL"],"title_canon_sha256":"5aaf4123d661aef6336c8f20328bbf04db19afa2e61f2d7ff7983edbaaf74ed5","abstract_canon_sha256":"13d47524bb81d0e5a173b75dbe4c9b7eb2730e8e59d5af842369367044a97240"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T07:46:02.161723Z","signature_b64":"p8Ck5AkppmGe/vH0eZ2XZtv2DifXsx+kpu2xtBuokTXwUVMwTFQ+9VyIGiaCSpN9phZuVwu0Tna/nHfuipgvAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"7860c54e4e00c9e269fa924f8a8802f589121963e30d080d10e948be641226c2","last_reissued_at":"2026-07-05T07:46:02.161286Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T07:46:02.161286Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"RLVF: Learning from Verbal Feedback without Overgeneralization","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.LG","authors_text":"Alexander Khazatsky, Annie S Chen, Archit Sharma, Chelsea Finn, Eric Mitchell, Moritz Stephan, Sheryl Hsu","submitted_at":"2024-02-16T18:50:24Z","abstract_excerpt":"The diversity of contexts in which large language models (LLMs) are deployed requires the ability to modify or customize default model behaviors to incorporate nuanced requirements and preferences. A convenient interface to specify such model adjustments is high-level verbal feedback, such as \"Don't use emojis when drafting emails to my boss.\" However, while writing high-level feedback is far simpler than collecting annotations for reinforcement learning from human feedback (RLHF), we find that simply prompting a model with such feedback leads to overgeneralization of the feedback to contexts "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2402.10893","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/2402.10893/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":"2402.10893","created_at":"2026-07-05T07:46:02.161356+00:00"},{"alias_kind":"arxiv_version","alias_value":"2402.10893v1","created_at":"2026-07-05T07:46:02.161356+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2402.10893","created_at":"2026-07-05T07:46:02.161356+00:00"},{"alias_kind":"pith_short_12","alias_value":"PBQMKTSOADE6","created_at":"2026-07-05T07:46:02.161356+00:00"},{"alias_kind":"pith_short_16","alias_value":"PBQMKTSOADE6E2P2","created_at":"2026-07-05T07:46:02.161356+00:00"},{"alias_kind":"pith_short_8","alias_value":"PBQMKTSO","created_at":"2026-07-05T07:46:02.161356+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2502.19417","citing_title":"Hi Robot: Open-Ended Instruction Following with Hierarchical Vision-Language-Action Models","ref_index":41,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/PBQMKTSOADE6E2P2SJHYVCAC6W","json":"https://pith.science/pith/PBQMKTSOADE6E2P2SJHYVCAC6W.json","graph_json":"https://pith.science/api/pith-number/PBQMKTSOADE6E2P2SJHYVCAC6W/graph.json","events_json":"https://pith.science/api/pith-number/PBQMKTSOADE6E2P2SJHYVCAC6W/events.json","paper":"https://pith.science/paper/PBQMKTSO"},"agent_actions":{"view_html":"https://pith.science/pith/PBQMKTSOADE6E2P2SJHYVCAC6W","download_json":"https://pith.science/pith/PBQMKTSOADE6E2P2SJHYVCAC6W.json","view_paper":"https://pith.science/paper/PBQMKTSO","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2402.10893&json=true","fetch_graph":"https://pith.science/api/pith-number/PBQMKTSOADE6E2P2SJHYVCAC6W/graph.json","fetch_events":"https://pith.science/api/pith-number/PBQMKTSOADE6E2P2SJHYVCAC6W/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/PBQMKTSOADE6E2P2SJHYVCAC6W/action/timestamp_anchor","attest_storage":"https://pith.science/pith/PBQMKTSOADE6E2P2SJHYVCAC6W/action/storage_attestation","attest_author":"https://pith.science/pith/PBQMKTSOADE6E2P2SJHYVCAC6W/action/author_attestation","sign_citation":"https://pith.science/pith/PBQMKTSOADE6E2P2SJHYVCAC6W/action/citation_signature","submit_replication":"https://pith.science/pith/PBQMKTSOADE6E2P2SJHYVCAC6W/action/replication_record"}},"created_at":"2026-07-05T07:46:02.161356+00:00","updated_at":"2026-07-05T07:46:02.161356+00:00"}