{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2021:FQLV5H64FSKHGJ2AKS5YY4EBGL","short_pith_number":"pith:FQLV5H64","schema_version":"1.0","canonical_sha256":"2c175e9fdc2c9473274054bb8c708132ca41b31dfe7f11ab065077eb6ab8d5fc","source":{"kind":"arxiv","id":"2107.09645","version":1},"attestation_state":"computed","paper":{"title":"Mastering Visual Continuous Control: Improved Data-Augmented Reinforcement Learning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.AI","authors_text":"Alessandro Lazaric, Denis Yarats, Lerrel Pinto, Rob Fergus","submitted_at":"2021-07-20T17:29:13Z","abstract_excerpt":"We present DrQ-v2, a model-free reinforcement learning (RL) algorithm for visual continuous control. DrQ-v2 builds on DrQ, an off-policy actor-critic approach that uses data augmentation to learn directly from pixels. We introduce several improvements that yield state-of-the-art results on the DeepMind Control Suite. Notably, DrQ-v2 is able to solve complex humanoid locomotion tasks directly from pixel observations, previously unattained by model-free RL. DrQ-v2 is conceptually simple, easy to implement, and provides significantly better computational footprint compared to prior work, with the"},"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":"2107.09645","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AI","submitted_at":"2021-07-20T17:29:13Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"74f758eb9658f72dd942327b1537a2085c5462da52f6addfe2e3154e13c42627","abstract_canon_sha256":"fe2f46042c94430ac584f5945b6229afc7e3e5bc40d4ee36ccda03e94972c32a"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T02:59:33.505206Z","signature_b64":"+m8BoaJocOVd2UY3fi/ujjYRNjBOpbyARq5RH6NWgzgjJmowgimT/Ek3hV0q9r4IirHgqqVRHc6ghoNP6wraDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"2c175e9fdc2c9473274054bb8c708132ca41b31dfe7f11ab065077eb6ab8d5fc","last_reissued_at":"2026-07-05T02:59:33.504683Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T02:59:33.504683Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Mastering Visual Continuous Control: Improved Data-Augmented Reinforcement Learning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.AI","authors_text":"Alessandro Lazaric, Denis Yarats, Lerrel Pinto, Rob Fergus","submitted_at":"2021-07-20T17:29:13Z","abstract_excerpt":"We present DrQ-v2, a model-free reinforcement learning (RL) algorithm for visual continuous control. DrQ-v2 builds on DrQ, an off-policy actor-critic approach that uses data augmentation to learn directly from pixels. We introduce several improvements that yield state-of-the-art results on the DeepMind Control Suite. Notably, DrQ-v2 is able to solve complex humanoid locomotion tasks directly from pixel observations, previously unattained by model-free RL. DrQ-v2 is conceptually simple, easy to implement, and provides significantly better computational footprint compared to prior work, with the"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2107.09645","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/2107.09645/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":"2107.09645","created_at":"2026-07-05T02:59:33.504749+00:00"},{"alias_kind":"arxiv_version","alias_value":"2107.09645v1","created_at":"2026-07-05T02:59:33.504749+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2107.09645","created_at":"2026-07-05T02:59:33.504749+00:00"},{"alias_kind":"pith_short_12","alias_value":"FQLV5H64FSKH","created_at":"2026-07-05T02:59:33.504749+00:00"},{"alias_kind":"pith_short_16","alias_value":"FQLV5H64FSKHGJ2A","created_at":"2026-07-05T02:59:33.504749+00:00"},{"alias_kind":"pith_short_8","alias_value":"FQLV5H64","created_at":"2026-07-05T02:59:33.504749+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":16,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.22027","citing_title":"RARM: Confidence-Gated Progress Reward Modeling for RL in Manipulation","ref_index":57,"is_internal_anchor":false},{"citing_arxiv_id":"2607.01897","citing_title":"Rank-Then-Act: Reward-Free Control from Frame-Order Progress","ref_index":27,"is_internal_anchor":false},{"citing_arxiv_id":"2605.26357","citing_title":"Balancing Plasticity and Stability with Fast and Slow Successor Features","ref_index":53,"is_internal_anchor":false},{"citing_arxiv_id":"2605.26478","citing_title":"Efficient On-policy Visual-RL via Stochastic Decoupled Policy Gradient","ref_index":1,"is_internal_anchor":false},{"citing_arxiv_id":"2310.02635","citing_title":"Reinforcement Learning with Foundation Priors: Let the Embodied Agent Efficiently Learn on Its Own","ref_index":11,"is_internal_anchor":false},{"citing_arxiv_id":"2509.26627","citing_title":"TimeRewarder: Learning Dense Reward from Passive Videos via Frame-wise Temporal Distance","ref_index":16,"is_internal_anchor":false},{"citing_arxiv_id":"2605.17336","citing_title":"Tactile-based Multimodal Fusion in Embodied Intelligence: A Survey of Vision, Language, and Contact-Driven Paradigms","ref_index":77,"is_internal_anchor":false},{"citing_arxiv_id":"2604.04539","citing_title":"FlashSAC: Fast and Stable Off-Policy Reinforcement Learning for High-Dimensional Robot Control","ref_index":92,"is_internal_anchor":false},{"citing_arxiv_id":"2509.16615","citing_title":"LLM-Guided Task- and Affordance-Level Exploration in Reinforcement Learning","ref_index":39,"is_internal_anchor":false},{"citing_arxiv_id":"2503.10631","citing_title":"HybridVLA: Collaborative Diffusion and Autoregression in a Unified Vision-Language-Action Model","ref_index":35,"is_internal_anchor":false},{"citing_arxiv_id":"2605.14350","citing_title":"Distributionally Robust Multi-Task Reinforcement Learning via Adaptive Task Sampling","ref_index":133,"is_internal_anchor":false},{"citing_arxiv_id":"2604.24661","citing_title":"Agent-Centric Observation Adaptation for Robust Visual Control under Dynamic Perturbations","ref_index":67,"is_internal_anchor":false},{"citing_arxiv_id":"2605.06500","citing_title":"Operator-Guided Invariance Learning for Continuous Reinforcement Learning","ref_index":4,"is_internal_anchor":false},{"citing_arxiv_id":"2301.04104","citing_title":"Mastering Diverse Domains through World Models","ref_index":47,"is_internal_anchor":false},{"citing_arxiv_id":"2604.24661","citing_title":"Agent-Centric Observation Adaptation for Robust Visual Control under Dynamic Perturbations","ref_index":67,"is_internal_anchor":false},{"citing_arxiv_id":"2604.04539","citing_title":"FlashSAC: Fast and Stable Off-Policy Reinforcement Learning for High-Dimensional Robot Control","ref_index":92,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/FQLV5H64FSKHGJ2AKS5YY4EBGL","json":"https://pith.science/pith/FQLV5H64FSKHGJ2AKS5YY4EBGL.json","graph_json":"https://pith.science/api/pith-number/FQLV5H64FSKHGJ2AKS5YY4EBGL/graph.json","events_json":"https://pith.science/api/pith-number/FQLV5H64FSKHGJ2AKS5YY4EBGL/events.json","paper":"https://pith.science/paper/FQLV5H64"},"agent_actions":{"view_html":"https://pith.science/pith/FQLV5H64FSKHGJ2AKS5YY4EBGL","download_json":"https://pith.science/pith/FQLV5H64FSKHGJ2AKS5YY4EBGL.json","view_paper":"https://pith.science/paper/FQLV5H64","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2107.09645&json=true","fetch_graph":"https://pith.science/api/pith-number/FQLV5H64FSKHGJ2AKS5YY4EBGL/graph.json","fetch_events":"https://pith.science/api/pith-number/FQLV5H64FSKHGJ2AKS5YY4EBGL/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/FQLV5H64FSKHGJ2AKS5YY4EBGL/action/timestamp_anchor","attest_storage":"https://pith.science/pith/FQLV5H64FSKHGJ2AKS5YY4EBGL/action/storage_attestation","attest_author":"https://pith.science/pith/FQLV5H64FSKHGJ2AKS5YY4EBGL/action/author_attestation","sign_citation":"https://pith.science/pith/FQLV5H64FSKHGJ2AKS5YY4EBGL/action/citation_signature","submit_replication":"https://pith.science/pith/FQLV5H64FSKHGJ2AKS5YY4EBGL/action/replication_record"}},"created_at":"2026-07-05T02:59:33.504749+00:00","updated_at":"2026-07-05T02:59:33.504749+00:00"}