{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2016:RSU7P2S2KWCBCBF7XXQCOATW4M","short_pith_number":"pith:RSU7P2S2","schema_version":"1.0","canonical_sha256":"8ca9f7ea5a55841104bfbde0270276e33ae40b778632e12d9a0beebdc5a83af7","source":{"kind":"arxiv","id":"1607.07086","version":3},"attestation_state":"computed","paper":{"title":"An Actor-Critic Algorithm for Sequence Prediction","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Aaron Courville, Anirudh Goyal, Dzmitry Bahdanau, Joelle Pineau, Kelvin Xu, Philemon Brakel, Ryan Lowe, Yoshua Bengio","submitted_at":"2016-07-24T20:05:07Z","abstract_excerpt":"We present an approach to training neural networks to generate sequences using actor-critic methods from reinforcement learning (RL). Current log-likelihood training methods are limited by the discrepancy between their training and testing modes, as models must generate tokens conditioned on their previous guesses rather than the ground-truth tokens. We address this problem by introducing a \\textit{critic} network that is trained to predict the value of an output token, given the policy of an \\textit{actor} network. This results in a training procedure that is much closer to the test phase, an"},"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":"1607.07086","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2016-07-24T20:05:07Z","cross_cats_sorted":[],"title_canon_sha256":"904e7816b10a1129b72a79dc1704fe413f5fe21230a8b0548df6d82f456f121b","abstract_canon_sha256":"d8fb0c6422698d85902586e2b27005611c470594ca2e4e6b545e4267fc430a4a"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-18T00:49:36.660112Z","signature_b64":"o8uAVVLOIQH8FgHyOD+3QLy4MqfVGZAlBex0JKA4ktGQv5oRsbDr3dXh6TUQSvXZrHSKZ2Ikh1swho5sXDwuDg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"8ca9f7ea5a55841104bfbde0270276e33ae40b778632e12d9a0beebdc5a83af7","last_reissued_at":"2026-05-18T00:49:36.659419Z","signature_status":"signed_v1","first_computed_at":"2026-05-18T00:49:36.659419Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"An Actor-Critic Algorithm for Sequence Prediction","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Aaron Courville, Anirudh Goyal, Dzmitry Bahdanau, Joelle Pineau, Kelvin Xu, Philemon Brakel, Ryan Lowe, Yoshua Bengio","submitted_at":"2016-07-24T20:05:07Z","abstract_excerpt":"We present an approach to training neural networks to generate sequences using actor-critic methods from reinforcement learning (RL). Current log-likelihood training methods are limited by the discrepancy between their training and testing modes, as models must generate tokens conditioned on their previous guesses rather than the ground-truth tokens. We address this problem by introducing a \\textit{critic} network that is trained to predict the value of an output token, given the policy of an \\textit{actor} network. This results in a training procedure that is much closer to the test phase, an"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"1607.07086","kind":"arxiv","version":3},"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":"1607.07086","created_at":"2026-05-18T00:49:36.659524+00:00"},{"alias_kind":"arxiv_version","alias_value":"1607.07086v3","created_at":"2026-05-18T00:49:36.659524+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1607.07086","created_at":"2026-05-18T00:49:36.659524+00:00"},{"alias_kind":"pith_short_12","alias_value":"RSU7P2S2KWCB","created_at":"2026-05-18T12:30:41.710351+00:00"},{"alias_kind":"pith_short_16","alias_value":"RSU7P2S2KWCBCBF7","created_at":"2026-05-18T12:30:41.710351+00:00"},{"alias_kind":"pith_short_8","alias_value":"RSU7P2S2","created_at":"2026-05-18T12:30:41.710351+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":6,"internal_anchor_count":5,"sample":[{"citing_arxiv_id":"1906.09444","citing_title":"Retrieving Sequential Information for Non-Autoregressive Neural Machine Translation","ref_index":1,"is_internal_anchor":true},{"citing_arxiv_id":"1907.11754","citing_title":"Deep Reinforcement Learning for Personalized Search Story Recommendation","ref_index":19,"is_internal_anchor":true},{"citing_arxiv_id":"2605.20999","citing_title":"Concentration of General Stochastic Approximation Under Heavy-Tailed Markovian Noise","ref_index":134,"is_internal_anchor":true},{"citing_arxiv_id":"2009.01325","citing_title":"Learning to summarize from human feedback","ref_index":1,"is_internal_anchor":true},{"citing_arxiv_id":"2302.12192","citing_title":"Aligning Text-to-Image Models using Human Feedback","ref_index":2,"is_internal_anchor":true},{"citing_arxiv_id":"2604.17184","citing_title":"SynthFix: Adaptive Neuro-Symbolic Code Vulnerability Repair","ref_index":65,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/RSU7P2S2KWCBCBF7XXQCOATW4M","json":"https://pith.science/pith/RSU7P2S2KWCBCBF7XXQCOATW4M.json","graph_json":"https://pith.science/api/pith-number/RSU7P2S2KWCBCBF7XXQCOATW4M/graph.json","events_json":"https://pith.science/api/pith-number/RSU7P2S2KWCBCBF7XXQCOATW4M/events.json","paper":"https://pith.science/paper/RSU7P2S2"},"agent_actions":{"view_html":"https://pith.science/pith/RSU7P2S2KWCBCBF7XXQCOATW4M","download_json":"https://pith.science/pith/RSU7P2S2KWCBCBF7XXQCOATW4M.json","view_paper":"https://pith.science/paper/RSU7P2S2","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=1607.07086&json=true","fetch_graph":"https://pith.science/api/pith-number/RSU7P2S2KWCBCBF7XXQCOATW4M/graph.json","fetch_events":"https://pith.science/api/pith-number/RSU7P2S2KWCBCBF7XXQCOATW4M/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/RSU7P2S2KWCBCBF7XXQCOATW4M/action/timestamp_anchor","attest_storage":"https://pith.science/pith/RSU7P2S2KWCBCBF7XXQCOATW4M/action/storage_attestation","attest_author":"https://pith.science/pith/RSU7P2S2KWCBCBF7XXQCOATW4M/action/author_attestation","sign_citation":"https://pith.science/pith/RSU7P2S2KWCBCBF7XXQCOATW4M/action/citation_signature","submit_replication":"https://pith.science/pith/RSU7P2S2KWCBCBF7XXQCOATW4M/action/replication_record"}},"created_at":"2026-05-18T00:49:36.659524+00:00","updated_at":"2026-05-18T00:49:36.659524+00:00"}