{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2019:MCQMVIWLO3R76TD7O5TOYH3OSU","short_pith_number":"pith:MCQMVIWL","schema_version":"1.0","canonical_sha256":"60a0caa2cb76e3ff4c7f7766ec1f6e951b49bd8dd6e1cb19ab5873bcd043b8fe","source":{"kind":"arxiv","id":"1907.00456","version":2},"attestation_state":"computed","paper":{"title":"Way Off-Policy Batch Deep Reinforcement Learning of Implicit Human Preferences in Dialog","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","stat.ML"],"primary_cat":"cs.LG","authors_text":"Agata Lapedriza, Asma Ghandeharioun, Craig Ferguson, Judy Hanwen Shen, Natasha Jaques, Noah Jones, Rosalind Picard, Shixiang Gu","submitted_at":"2019-06-30T20:53:19Z","abstract_excerpt":"Most deep reinforcement learning (RL) systems are not able to learn effectively from off-policy data, especially if they cannot explore online in the environment. These are critical shortcomings for applying RL to real-world problems where collecting data is expensive, and models must be tested offline before being deployed to interact with the environment -- e.g. systems that learn from human interaction. Thus, we develop a novel class of off-policy batch RL algorithms, which are able to effectively learn offline, without exploring, from a fixed batch of human interaction data. We leverage mo"},"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":"1907.00456","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2019-06-30T20:53:19Z","cross_cats_sorted":["cs.AI","stat.ML"],"title_canon_sha256":"edb9f11dda92f2c090f26a59390efb333c67c887e01c84879fc1797f4815041d","abstract_canon_sha256":"05b5d99512a3744648aac17c70914a96cdc29a1128bfa051a1171409f9148b82"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-17T23:41:17.029737Z","signature_b64":"M+fya589XvJpMgzL+pfQC5+FaVM0I9tNe2I+Hs8DaNRHaoU3U55DlFDMYVxg+aA0Lgan78scTcx01BYRohtCCQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"60a0caa2cb76e3ff4c7f7766ec1f6e951b49bd8dd6e1cb19ab5873bcd043b8fe","last_reissued_at":"2026-05-17T23:41:17.029130Z","signature_status":"signed_v1","first_computed_at":"2026-05-17T23:41:17.029130Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Way Off-Policy Batch Deep Reinforcement Learning of Implicit Human Preferences in Dialog","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","stat.ML"],"primary_cat":"cs.LG","authors_text":"Agata Lapedriza, Asma Ghandeharioun, Craig Ferguson, Judy Hanwen Shen, Natasha Jaques, Noah Jones, Rosalind Picard, Shixiang Gu","submitted_at":"2019-06-30T20:53:19Z","abstract_excerpt":"Most deep reinforcement learning (RL) systems are not able to learn effectively from off-policy data, especially if they cannot explore online in the environment. These are critical shortcomings for applying RL to real-world problems where collecting data is expensive, and models must be tested offline before being deployed to interact with the environment -- e.g. systems that learn from human interaction. Thus, we develop a novel class of off-policy batch RL algorithms, which are able to effectively learn offline, without exploring, from a fixed batch of human interaction data. We leverage mo"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"1907.00456","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":""},"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":"1907.00456","created_at":"2026-05-17T23:41:17.029228+00:00"},{"alias_kind":"arxiv_version","alias_value":"1907.00456v2","created_at":"2026-05-17T23:41:17.029228+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1907.00456","created_at":"2026-05-17T23:41:17.029228+00:00"},{"alias_kind":"pith_short_12","alias_value":"MCQMVIWLO3R7","created_at":"2026-05-18T12:33:21.387695+00:00"},{"alias_kind":"pith_short_16","alias_value":"MCQMVIWLO3R76TD7","created_at":"2026-05-18T12:33:21.387695+00:00"},{"alias_kind":"pith_short_8","alias_value":"MCQMVIWL","created_at":"2026-05-18T12:33:21.387695+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":15,"internal_anchor_count":6,"sample":[{"citing_arxiv_id":"2507.10797","citing_title":"Multi-Armed Sampling Problem and the End of Exploration","ref_index":14,"is_internal_anchor":true},{"citing_arxiv_id":"2009.01325","citing_title":"Learning to summarize from human feedback","ref_index":25,"is_internal_anchor":true},{"citing_arxiv_id":"1910.01708","citing_title":"Benchmarking Batch Deep Reinforcement Learning Algorithms","ref_index":9,"is_internal_anchor":true},{"citing_arxiv_id":"2406.10162","citing_title":"Sycophancy to Subterfuge: Investigating Reward-Tampering in Large Language Models","ref_index":186,"is_internal_anchor":true},{"citing_arxiv_id":"2307.04964","citing_title":"Secrets of RLHF in Large Language Models Part I: PPO","ref_index":32,"is_internal_anchor":true},{"citing_arxiv_id":"2605.14350","citing_title":"Distributionally Robust Multi-Task Reinforcement Learning via Adaptive Task Sampling","ref_index":20,"is_internal_anchor":true},{"citing_arxiv_id":"1911.11361","citing_title":"Behavior Regularized Offline Reinforcement Learning","ref_index":10,"is_internal_anchor":false},{"citing_arxiv_id":"2605.12416","citing_title":"Aligning Flow Map Policies with Optimal Q-Guidance","ref_index":17,"is_internal_anchor":false},{"citing_arxiv_id":"2006.09359","citing_title":"AWAC: Accelerating Online Reinforcement Learning with Offline Datasets","ref_index":23,"is_internal_anchor":false},{"citing_arxiv_id":"2004.07219","citing_title":"D4RL: Datasets for Deep Data-Driven Reinforcement Learning","ref_index":13,"is_internal_anchor":false},{"citing_arxiv_id":"2202.03286","citing_title":"Red Teaming Language Models with Language Models","ref_index":4,"is_internal_anchor":false},{"citing_arxiv_id":"2605.06529","citing_title":"Market-Alignment Risk in Pricing Agents: Trace Diagnostics and Trace-Prior RL under Hidden Competitor State","ref_index":4,"is_internal_anchor":false},{"citing_arxiv_id":"2605.04653","citing_title":"Threshold-Guided Optimization for Visual Generative Models","ref_index":42,"is_internal_anchor":false},{"citing_arxiv_id":"2005.01643","citing_title":"Offline Reinforcement Learning: Tutorial, Review, and Perspectives on Open Problems","ref_index":206,"is_internal_anchor":false},{"citing_arxiv_id":"1909.08593","citing_title":"Fine-Tuning Language Models from Human Preferences","ref_index":11,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/MCQMVIWLO3R76TD7O5TOYH3OSU","json":"https://pith.science/pith/MCQMVIWLO3R76TD7O5TOYH3OSU.json","graph_json":"https://pith.science/api/pith-number/MCQMVIWLO3R76TD7O5TOYH3OSU/graph.json","events_json":"https://pith.science/api/pith-number/MCQMVIWLO3R76TD7O5TOYH3OSU/events.json","paper":"https://pith.science/paper/MCQMVIWL"},"agent_actions":{"view_html":"https://pith.science/pith/MCQMVIWLO3R76TD7O5TOYH3OSU","download_json":"https://pith.science/pith/MCQMVIWLO3R76TD7O5TOYH3OSU.json","view_paper":"https://pith.science/paper/MCQMVIWL","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=1907.00456&json=true","fetch_graph":"https://pith.science/api/pith-number/MCQMVIWLO3R76TD7O5TOYH3OSU/graph.json","fetch_events":"https://pith.science/api/pith-number/MCQMVIWLO3R76TD7O5TOYH3OSU/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/MCQMVIWLO3R76TD7O5TOYH3OSU/action/timestamp_anchor","attest_storage":"https://pith.science/pith/MCQMVIWLO3R76TD7O5TOYH3OSU/action/storage_attestation","attest_author":"https://pith.science/pith/MCQMVIWLO3R76TD7O5TOYH3OSU/action/author_attestation","sign_citation":"https://pith.science/pith/MCQMVIWLO3R76TD7O5TOYH3OSU/action/citation_signature","submit_replication":"https://pith.science/pith/MCQMVIWLO3R76TD7O5TOYH3OSU/action/replication_record"}},"created_at":"2026-05-17T23:41:17.029228+00:00","updated_at":"2026-05-17T23:41:17.029228+00:00"}