{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:TGBTB2W2W4WC6BBYB2AM6MU72M","short_pith_number":"pith:TGBTB2W2","schema_version":"1.0","canonical_sha256":"998330eadab72c2f04380e80cf329fd320e191d4984bfa9fa01d917cbfdf4d2c","source":{"kind":"arxiv","id":"2505.23150","version":1},"attestation_state":"computed","paper":{"title":"Bigger, Regularized, Categorical: High-Capacity Value Functions are Efficient Multi-Task Learners","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Aviral Kumar, Carmelo Sferrazza, Marek Cygan, Michal Nauman, Pieter Abbeel","submitted_at":"2025-05-29T06:41:45Z","abstract_excerpt":"Recent advances in language modeling and vision stem from training large models on diverse, multi-task data. This paradigm has had limited impact in value-based reinforcement learning (RL), where improvements are often driven by small models trained in a single-task context. This is because in multi-task RL sparse rewards and gradient conflicts make optimization of temporal difference brittle. Practical workflows for generalist policies therefore avoid online training, instead cloning expert trajectories or distilling collections of single-task policies into one agent. In this work, we show th"},"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":"2505.23150","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-05-29T06:41:45Z","cross_cats_sorted":[],"title_canon_sha256":"93c7a6d8e9fde778087dcc5a4b572671f18b6961fe8983072712275e03b7f873","abstract_canon_sha256":"0827efdcd92357dc2623bccb295810cfd21885526ab1031377e04c20d5f373f3"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:11:56.342674Z","signature_b64":"A6K2zAQ0E0/IMzX0Ij3qlKJk3YxgW7XLi1/TvveUtdY2ePRs/wX74Zgo0avtWUtbg6avs+O2gkLCcUMuHZvrBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"998330eadab72c2f04380e80cf329fd320e191d4984bfa9fa01d917cbfdf4d2c","last_reissued_at":"2026-07-05T11:11:56.342217Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:11:56.342217Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Bigger, Regularized, Categorical: High-Capacity Value Functions are Efficient Multi-Task Learners","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Aviral Kumar, Carmelo Sferrazza, Marek Cygan, Michal Nauman, Pieter Abbeel","submitted_at":"2025-05-29T06:41:45Z","abstract_excerpt":"Recent advances in language modeling and vision stem from training large models on diverse, multi-task data. This paradigm has had limited impact in value-based reinforcement learning (RL), where improvements are often driven by small models trained in a single-task context. This is because in multi-task RL sparse rewards and gradient conflicts make optimization of temporal difference brittle. Practical workflows for generalist policies therefore avoid online training, instead cloning expert trajectories or distilling collections of single-task policies into one agent. In this work, we show th"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2505.23150","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/2505.23150/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":"2505.23150","created_at":"2026-07-05T11:11:56.342278+00:00"},{"alias_kind":"arxiv_version","alias_value":"2505.23150v1","created_at":"2026-07-05T11:11:56.342278+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2505.23150","created_at":"2026-07-05T11:11:56.342278+00:00"},{"alias_kind":"pith_short_12","alias_value":"TGBTB2W2W4WC","created_at":"2026-07-05T11:11:56.342278+00:00"},{"alias_kind":"pith_short_16","alias_value":"TGBTB2W2W4WC6BBY","created_at":"2026-07-05T11:11:56.342278+00:00"},{"alias_kind":"pith_short_8","alias_value":"TGBTB2W2","created_at":"2026-07-05T11:11:56.342278+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":7,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.10236","citing_title":"When Does Non-Uniform Replay Matter in Reinforcement Learning?","ref_index":23,"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":61,"is_internal_anchor":false},{"citing_arxiv_id":"2603.04333","citing_title":"What Does Flow Matching Bring To TD Learning?","ref_index":43,"is_internal_anchor":false},{"citing_arxiv_id":"2605.11711","citing_title":"Debiased Model-based Representations for Sample-efficient Continuous Control","ref_index":10,"is_internal_anchor":false},{"citing_arxiv_id":"2605.10236","citing_title":"When Does Non-Uniform Replay Matter in Reinforcement Learning?","ref_index":23,"is_internal_anchor":false},{"citing_arxiv_id":"2605.10236","citing_title":"When Does Non-Uniform Replay Matter in Reinforcement Learning?","ref_index":23,"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":61,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/TGBTB2W2W4WC6BBYB2AM6MU72M","json":"https://pith.science/pith/TGBTB2W2W4WC6BBYB2AM6MU72M.json","graph_json":"https://pith.science/api/pith-number/TGBTB2W2W4WC6BBYB2AM6MU72M/graph.json","events_json":"https://pith.science/api/pith-number/TGBTB2W2W4WC6BBYB2AM6MU72M/events.json","paper":"https://pith.science/paper/TGBTB2W2"},"agent_actions":{"view_html":"https://pith.science/pith/TGBTB2W2W4WC6BBYB2AM6MU72M","download_json":"https://pith.science/pith/TGBTB2W2W4WC6BBYB2AM6MU72M.json","view_paper":"https://pith.science/paper/TGBTB2W2","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2505.23150&json=true","fetch_graph":"https://pith.science/api/pith-number/TGBTB2W2W4WC6BBYB2AM6MU72M/graph.json","fetch_events":"https://pith.science/api/pith-number/TGBTB2W2W4WC6BBYB2AM6MU72M/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/TGBTB2W2W4WC6BBYB2AM6MU72M/action/timestamp_anchor","attest_storage":"https://pith.science/pith/TGBTB2W2W4WC6BBYB2AM6MU72M/action/storage_attestation","attest_author":"https://pith.science/pith/TGBTB2W2W4WC6BBYB2AM6MU72M/action/author_attestation","sign_citation":"https://pith.science/pith/TGBTB2W2W4WC6BBYB2AM6MU72M/action/citation_signature","submit_replication":"https://pith.science/pith/TGBTB2W2W4WC6BBYB2AM6MU72M/action/replication_record"}},"created_at":"2026-07-05T11:11:56.342278+00:00","updated_at":"2026-07-05T11:11:56.342278+00:00"}