{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2020:3BQWZ52JXPWGXO5WYSI2RITLWK","short_pith_number":"pith:3BQWZ52J","schema_version":"1.0","canonical_sha256":"d8616cf749bbec6bbbb6c491a8a26bb283be6aded42bb9e79875c02fde160b77","source":{"kind":"arxiv","id":"2001.11113","version":7},"attestation_state":"computed","paper":{"title":"GradientDICE: Rethinking Generalized Offline Estimation of Stationary Values","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["stat.ML"],"primary_cat":"cs.LG","authors_text":"Bo Liu, Shangtong Zhang, Shimon Whiteson","submitted_at":"2020-01-29T22:10:11Z","abstract_excerpt":"We present GradientDICE for estimating the density ratio between the state distribution of the target policy and the sampling distribution in off-policy reinforcement learning. GradientDICE fixes several problems of GenDICE (Zhang et al., 2020), the state-of-the-art for estimating such density ratios. Namely, the optimization problem in GenDICE is not a convex-concave saddle-point problem once nonlinearity in optimization variable parameterization is introduced to ensure positivity, so any primal-dual algorithm is not guaranteed to converge or find the desired solution. However, such nonlinear"},"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":"2001.11113","kind":"arxiv","version":7},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2020-01-29T22:10:11Z","cross_cats_sorted":["stat.ML"],"title_canon_sha256":"0557ddbe1f980289245498e81c630ada919021fff4e9154be74d6bf535897f53","abstract_canon_sha256":"8c2c04c78f505aaf54ce72cf57e7e730508447bedb749e4f4fb6076ef5b7bfee"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T01:54:42.595884Z","signature_b64":"6e4IbFCTvY3y/qsWu5Dtg1TFixpG22jk4tO2vAE8YrUEFfm8y6f+1vhvX/SGQEq7WXQHb3YwnxQCn4oThG1dDw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"d8616cf749bbec6bbbb6c491a8a26bb283be6aded42bb9e79875c02fde160b77","last_reissued_at":"2026-07-05T01:54:42.595457Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T01:54:42.595457Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"GradientDICE: Rethinking Generalized Offline Estimation of Stationary Values","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["stat.ML"],"primary_cat":"cs.LG","authors_text":"Bo Liu, Shangtong Zhang, Shimon Whiteson","submitted_at":"2020-01-29T22:10:11Z","abstract_excerpt":"We present GradientDICE for estimating the density ratio between the state distribution of the target policy and the sampling distribution in off-policy reinforcement learning. GradientDICE fixes several problems of GenDICE (Zhang et al., 2020), the state-of-the-art for estimating such density ratios. Namely, the optimization problem in GenDICE is not a convex-concave saddle-point problem once nonlinearity in optimization variable parameterization is introduced to ensure positivity, so any primal-dual algorithm is not guaranteed to converge or find the desired solution. However, such nonlinear"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2001.11113","kind":"arxiv","version":7},"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/2001.11113/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":"2001.11113","created_at":"2026-07-05T01:54:42.595516+00:00"},{"alias_kind":"arxiv_version","alias_value":"2001.11113v7","created_at":"2026-07-05T01:54:42.595516+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2001.11113","created_at":"2026-07-05T01:54:42.595516+00:00"},{"alias_kind":"pith_short_12","alias_value":"3BQWZ52JXPWG","created_at":"2026-07-05T01:54:42.595516+00:00"},{"alias_kind":"pith_short_16","alias_value":"3BQWZ52JXPWGXO5W","created_at":"2026-07-05T01:54:42.595516+00:00"},{"alias_kind":"pith_short_8","alias_value":"3BQWZ52J","created_at":"2026-07-05T01:54:42.595516+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2005.01643","citing_title":"Offline Reinforcement Learning: Tutorial, Review, and Perspectives on Open Problems","ref_index":271,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/3BQWZ52JXPWGXO5WYSI2RITLWK","json":"https://pith.science/pith/3BQWZ52JXPWGXO5WYSI2RITLWK.json","graph_json":"https://pith.science/api/pith-number/3BQWZ52JXPWGXO5WYSI2RITLWK/graph.json","events_json":"https://pith.science/api/pith-number/3BQWZ52JXPWGXO5WYSI2RITLWK/events.json","paper":"https://pith.science/paper/3BQWZ52J"},"agent_actions":{"view_html":"https://pith.science/pith/3BQWZ52JXPWGXO5WYSI2RITLWK","download_json":"https://pith.science/pith/3BQWZ52JXPWGXO5WYSI2RITLWK.json","view_paper":"https://pith.science/paper/3BQWZ52J","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2001.11113&json=true","fetch_graph":"https://pith.science/api/pith-number/3BQWZ52JXPWGXO5WYSI2RITLWK/graph.json","fetch_events":"https://pith.science/api/pith-number/3BQWZ52JXPWGXO5WYSI2RITLWK/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/3BQWZ52JXPWGXO5WYSI2RITLWK/action/timestamp_anchor","attest_storage":"https://pith.science/pith/3BQWZ52JXPWGXO5WYSI2RITLWK/action/storage_attestation","attest_author":"https://pith.science/pith/3BQWZ52JXPWGXO5WYSI2RITLWK/action/author_attestation","sign_citation":"https://pith.science/pith/3BQWZ52JXPWGXO5WYSI2RITLWK/action/citation_signature","submit_replication":"https://pith.science/pith/3BQWZ52JXPWGXO5WYSI2RITLWK/action/replication_record"}},"created_at":"2026-07-05T01:54:42.595516+00:00","updated_at":"2026-07-05T01:54:42.595516+00:00"}