{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2017:CFKYWVYHVCUMSCO5IMJAX3W2LO","short_pith_number":"pith:CFKYWVYH","schema_version":"1.0","canonical_sha256":"11558b5707a8a8c909dd43120beeda5ba5455946d447f627df34e0bbe39ea46c","source":{"kind":"arxiv","id":"1704.04651","version":2},"attestation_state":"computed","paper":{"title":"The Reactor: A fast and sample-efficient Actor-Critic agent for Reinforcement Learning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Audrunas Gruslys, Bilal Piot, Marc Bellemare, Mohammad Gheshlaghi Azar, Remi Munos, Will Dabney","submitted_at":"2017-04-15T15:38:23Z","abstract_excerpt":"In this work we present a new agent architecture, called Reactor, which combines multiple algorithmic and architectural contributions to produce an agent with higher sample-efficiency than Prioritized Dueling DQN (Wang et al., 2016) and Categorical DQN (Bellemare et al., 2017), while giving better run-time performance than A3C (Mnih et al., 2016). Our first contribution is a new policy evaluation algorithm called Distributional Retrace, which brings multi-step off-policy updates to the distributional reinforcement learning setting. The same approach can be used to convert several classes of mu"},"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":"1704.04651","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AI","submitted_at":"2017-04-15T15:38:23Z","cross_cats_sorted":[],"title_canon_sha256":"4c9d726b1a2108d46adf005994134ecfe8fcc60e08dd5199af1855b9e2db7bda","abstract_canon_sha256":"cca3cbd0158890019c41c410eb2b4d34a94f4ae2a0945835be264c24d450457c"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-18T00:12:56.951659Z","signature_b64":"mGbZMe6s73xqcqMsnXRlowhGakDy+kD5qeYyMDjBpzXGGvHWPnJTmRdWyYunjKprWlVEhfLlfXD1NWhg5k5RDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"11558b5707a8a8c909dd43120beeda5ba5455946d447f627df34e0bbe39ea46c","last_reissued_at":"2026-05-18T00:12:56.950919Z","signature_status":"signed_v1","first_computed_at":"2026-05-18T00:12:56.950919Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"The Reactor: A fast and sample-efficient Actor-Critic agent for Reinforcement Learning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Audrunas Gruslys, Bilal Piot, Marc Bellemare, Mohammad Gheshlaghi Azar, Remi Munos, Will Dabney","submitted_at":"2017-04-15T15:38:23Z","abstract_excerpt":"In this work we present a new agent architecture, called Reactor, which combines multiple algorithmic and architectural contributions to produce an agent with higher sample-efficiency than Prioritized Dueling DQN (Wang et al., 2016) and Categorical DQN (Bellemare et al., 2017), while giving better run-time performance than A3C (Mnih et al., 2016). Our first contribution is a new policy evaluation algorithm called Distributional Retrace, which brings multi-step off-policy updates to the distributional reinforcement learning setting. The same approach can be used to convert several classes of mu"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"1704.04651","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":"1704.04651","created_at":"2026-05-18T00:12:56.951051+00:00"},{"alias_kind":"arxiv_version","alias_value":"1704.04651v2","created_at":"2026-05-18T00:12:56.951051+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1704.04651","created_at":"2026-05-18T00:12:56.951051+00:00"},{"alias_kind":"pith_short_12","alias_value":"CFKYWVYHVCUM","created_at":"2026-05-18T12:31:10.602751+00:00"},{"alias_kind":"pith_short_16","alias_value":"CFKYWVYHVCUMSCO5","created_at":"2026-05-18T12:31:10.602751+00:00"},{"alias_kind":"pith_short_8","alias_value":"CFKYWVYH","created_at":"2026-05-18T12:31:10.602751+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":4,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2010.02193","citing_title":"Mastering Atari with Discrete World Models","ref_index":21,"is_internal_anchor":true},{"citing_arxiv_id":"1812.05905","citing_title":"Soft Actor-Critic Algorithms and Applications","ref_index":4,"is_internal_anchor":false},{"citing_arxiv_id":"1801.01290","citing_title":"Soft Actor-Critic: Off-Policy Maximum Entropy Deep Reinforcement Learning with a Stochastic Actor","ref_index":7,"is_internal_anchor":false},{"citing_arxiv_id":"2301.04104","citing_title":"Mastering Diverse Domains through World Models","ref_index":62,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/CFKYWVYHVCUMSCO5IMJAX3W2LO","json":"https://pith.science/pith/CFKYWVYHVCUMSCO5IMJAX3W2LO.json","graph_json":"https://pith.science/api/pith-number/CFKYWVYHVCUMSCO5IMJAX3W2LO/graph.json","events_json":"https://pith.science/api/pith-number/CFKYWVYHVCUMSCO5IMJAX3W2LO/events.json","paper":"https://pith.science/paper/CFKYWVYH"},"agent_actions":{"view_html":"https://pith.science/pith/CFKYWVYHVCUMSCO5IMJAX3W2LO","download_json":"https://pith.science/pith/CFKYWVYHVCUMSCO5IMJAX3W2LO.json","view_paper":"https://pith.science/paper/CFKYWVYH","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=1704.04651&json=true","fetch_graph":"https://pith.science/api/pith-number/CFKYWVYHVCUMSCO5IMJAX3W2LO/graph.json","fetch_events":"https://pith.science/api/pith-number/CFKYWVYHVCUMSCO5IMJAX3W2LO/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/CFKYWVYHVCUMSCO5IMJAX3W2LO/action/timestamp_anchor","attest_storage":"https://pith.science/pith/CFKYWVYHVCUMSCO5IMJAX3W2LO/action/storage_attestation","attest_author":"https://pith.science/pith/CFKYWVYHVCUMSCO5IMJAX3W2LO/action/author_attestation","sign_citation":"https://pith.science/pith/CFKYWVYHVCUMSCO5IMJAX3W2LO/action/citation_signature","submit_replication":"https://pith.science/pith/CFKYWVYHVCUMSCO5IMJAX3W2LO/action/replication_record"}},"created_at":"2026-05-18T00:12:56.951051+00:00","updated_at":"2026-05-18T00:12:56.951051+00:00"}