{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:5IRK7JBGRJNGJL5EVHWXVFNAGA","short_pith_number":"pith:5IRK7JBG","schema_version":"1.0","canonical_sha256":"ea22afa4268a5a64afa4a9ed7a95a03015a3f3e0173b690515b12e84f6ceb77f","source":{"kind":"arxiv","id":"2302.14372","version":2},"attestation_state":"computed","paper":{"title":"The In-Sample Softmax for Offline Reinforcement Learning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Adam White, Chenjun Xiao, Han Wang, Martha White, Yangchen Pan","submitted_at":"2023-02-28T07:55:02Z","abstract_excerpt":"Reinforcement learning (RL) agents can leverage batches of previously collected data to extract a reasonable control policy. An emerging issue in this offline RL setting, however, is that the bootstrapping update underlying many of our methods suffers from insufficient action-coverage: standard max operator may select a maximal action that has not been seen in the dataset. Bootstrapping from these inaccurate values can lead to overestimation and even divergence. There are a growing number of methods that attempt to approximate an \\emph{in-sample} max, that only uses actions well-covered by 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":"2302.14372","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2023-02-28T07:55:02Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"356cffcc511dae48aa8156d8c9a5c2abd48d5c7bdeb39d438d083ecff167cf64","abstract_canon_sha256":"bd24584cc858eeba8fda97992f2007e5d4ee81f1bb6413408478b8353961612c"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T06:02:32.599267Z","signature_b64":"/xToBuTgMZQI4qhTcLLpc52bbgfhwJc5F+chqUBOvZeOVOJ7BpVEcicTigHXEe5c5HRsbDwWW/gKwEpblEsABQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"ea22afa4268a5a64afa4a9ed7a95a03015a3f3e0173b690515b12e84f6ceb77f","last_reissued_at":"2026-07-05T06:02:32.598858Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T06:02:32.598858Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"The In-Sample Softmax for Offline Reinforcement Learning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Adam White, Chenjun Xiao, Han Wang, Martha White, Yangchen Pan","submitted_at":"2023-02-28T07:55:02Z","abstract_excerpt":"Reinforcement learning (RL) agents can leverage batches of previously collected data to extract a reasonable control policy. An emerging issue in this offline RL setting, however, is that the bootstrapping update underlying many of our methods suffers from insufficient action-coverage: standard max operator may select a maximal action that has not been seen in the dataset. Bootstrapping from these inaccurate values can lead to overestimation and even divergence. There are a growing number of methods that attempt to approximate an \\emph{in-sample} max, that only uses actions well-covered by the"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2302.14372","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":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2302.14372/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":"2302.14372","created_at":"2026-07-05T06:02:32.598917+00:00"},{"alias_kind":"arxiv_version","alias_value":"2302.14372v2","created_at":"2026-07-05T06:02:32.598917+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2302.14372","created_at":"2026-07-05T06:02:32.598917+00:00"},{"alias_kind":"pith_short_12","alias_value":"5IRK7JBGRJNG","created_at":"2026-07-05T06:02:32.598917+00:00"},{"alias_kind":"pith_short_16","alias_value":"5IRK7JBGRJNGJL5E","created_at":"2026-07-05T06:02:32.598917+00:00"},{"alias_kind":"pith_short_8","alias_value":"5IRK7JBG","created_at":"2026-07-05T06:02:32.598917+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.14779","citing_title":"Peng's Q($\\lambda$) for Conservative Value Estimation in Offline Reinforcement Learning","ref_index":17,"is_internal_anchor":false},{"citing_arxiv_id":"2605.27877","citing_title":"SPAR: Support-Preserving Action Rectification","ref_index":17,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/5IRK7JBGRJNGJL5EVHWXVFNAGA","json":"https://pith.science/pith/5IRK7JBGRJNGJL5EVHWXVFNAGA.json","graph_json":"https://pith.science/api/pith-number/5IRK7JBGRJNGJL5EVHWXVFNAGA/graph.json","events_json":"https://pith.science/api/pith-number/5IRK7JBGRJNGJL5EVHWXVFNAGA/events.json","paper":"https://pith.science/paper/5IRK7JBG"},"agent_actions":{"view_html":"https://pith.science/pith/5IRK7JBGRJNGJL5EVHWXVFNAGA","download_json":"https://pith.science/pith/5IRK7JBGRJNGJL5EVHWXVFNAGA.json","view_paper":"https://pith.science/paper/5IRK7JBG","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2302.14372&json=true","fetch_graph":"https://pith.science/api/pith-number/5IRK7JBGRJNGJL5EVHWXVFNAGA/graph.json","fetch_events":"https://pith.science/api/pith-number/5IRK7JBGRJNGJL5EVHWXVFNAGA/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/5IRK7JBGRJNGJL5EVHWXVFNAGA/action/timestamp_anchor","attest_storage":"https://pith.science/pith/5IRK7JBGRJNGJL5EVHWXVFNAGA/action/storage_attestation","attest_author":"https://pith.science/pith/5IRK7JBGRJNGJL5EVHWXVFNAGA/action/author_attestation","sign_citation":"https://pith.science/pith/5IRK7JBGRJNGJL5EVHWXVFNAGA/action/citation_signature","submit_replication":"https://pith.science/pith/5IRK7JBGRJNGJL5EVHWXVFNAGA/action/replication_record"}},"created_at":"2026-07-05T06:02:32.598917+00:00","updated_at":"2026-07-05T06:02:32.598917+00:00"}