{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:M4GQN4NWJTMQ3QWP6SZGOTHWZC","short_pith_number":"pith:M4GQN4NW","schema_version":"1.0","canonical_sha256":"670d06f1b64cd90dc2cff4b2674cf6c88e41da006c2f64ddc36ec7606e1c9411","source":{"kind":"arxiv","id":"2602.05459","version":2},"attestation_state":"computed","paper":{"title":"Beyond Success Rates: Trainability and Extractability for Offline GCRL","license":"http://creativecommons.org/licenses/by-sa/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Aditya Mohan, Jan Malte T\\\"opperwien, Marius Lindauer","submitted_at":"2026-02-05T09:08:17Z","abstract_excerpt":"Offline goal-conditioned reinforcement learning (GCRL) is typically benchmarked by the best tuned success rate of each method. This score measures attainable performance, but it does not reveal how reliably a learned goal-conditioned signal can be extracted into a policy: a method could succeed across many value-learning and extraction settings, or only at a narrow, hard-to-find configuration. We study this gap across four methods, GCIQL, GCIVL, QRL, and CRL, under a shared advantage-weighted regression (AWR) extractor. For each method, we construct trainability landscapes over the optimizer l"},"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":"2602.05459","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.LG","submitted_at":"2026-02-05T09:08:17Z","cross_cats_sorted":[],"title_canon_sha256":"cf8752ffbe4461e6ed19cc946cba45e34bf3b0d4103481744b732e4721e0fe3c","abstract_canon_sha256":"2b75102d5f3e0d39deba8f421c9ac72393149c679d2c83190751e1d9d505a6fe"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-10T01:19:42.041610Z","signature_b64":"z0vYp8VKvavrp3FHuXgud8I7Yy/OGKDTWf7zUiP2O/8lnFNjVjocXERfXjhYil5uHHz0CsxyrE30AuOJ+IL9BA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"670d06f1b64cd90dc2cff4b2674cf6c88e41da006c2f64ddc36ec7606e1c9411","last_reissued_at":"2026-07-10T01:19:42.041092Z","signature_status":"signed_v1","first_computed_at":"2026-07-10T01:19:42.041092Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Beyond Success Rates: Trainability and Extractability for Offline GCRL","license":"http://creativecommons.org/licenses/by-sa/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Aditya Mohan, Jan Malte T\\\"opperwien, Marius Lindauer","submitted_at":"2026-02-05T09:08:17Z","abstract_excerpt":"Offline goal-conditioned reinforcement learning (GCRL) is typically benchmarked by the best tuned success rate of each method. This score measures attainable performance, but it does not reveal how reliably a learned goal-conditioned signal can be extracted into a policy: a method could succeed across many value-learning and extraction settings, or only at a narrow, hard-to-find configuration. We study this gap across four methods, GCIQL, GCIVL, QRL, and CRL, under a shared advantage-weighted regression (AWR) extractor. For each method, we construct trainability landscapes over the optimizer l"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2602.05459","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/2602.05459/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":"2602.05459","created_at":"2026-07-10T01:19:42.041145+00:00"},{"alias_kind":"arxiv_version","alias_value":"2602.05459v2","created_at":"2026-07-10T01:19:42.041145+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2602.05459","created_at":"2026-07-10T01:19:42.041145+00:00"},{"alias_kind":"pith_short_12","alias_value":"M4GQN4NWJTMQ","created_at":"2026-07-10T01:19:42.041145+00:00"},{"alias_kind":"pith_short_16","alias_value":"M4GQN4NWJTMQ3QWP","created_at":"2026-07-10T01:19:42.041145+00:00"},{"alias_kind":"pith_short_8","alias_value":"M4GQN4NW","created_at":"2026-07-10T01:19:42.041145+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/M4GQN4NWJTMQ3QWP6SZGOTHWZC","json":"https://pith.science/pith/M4GQN4NWJTMQ3QWP6SZGOTHWZC.json","graph_json":"https://pith.science/api/pith-number/M4GQN4NWJTMQ3QWP6SZGOTHWZC/graph.json","events_json":"https://pith.science/api/pith-number/M4GQN4NWJTMQ3QWP6SZGOTHWZC/events.json","paper":"https://pith.science/paper/M4GQN4NW"},"agent_actions":{"view_html":"https://pith.science/pith/M4GQN4NWJTMQ3QWP6SZGOTHWZC","download_json":"https://pith.science/pith/M4GQN4NWJTMQ3QWP6SZGOTHWZC.json","view_paper":"https://pith.science/paper/M4GQN4NW","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2602.05459&json=true","fetch_graph":"https://pith.science/api/pith-number/M4GQN4NWJTMQ3QWP6SZGOTHWZC/graph.json","fetch_events":"https://pith.science/api/pith-number/M4GQN4NWJTMQ3QWP6SZGOTHWZC/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/M4GQN4NWJTMQ3QWP6SZGOTHWZC/action/timestamp_anchor","attest_storage":"https://pith.science/pith/M4GQN4NWJTMQ3QWP6SZGOTHWZC/action/storage_attestation","attest_author":"https://pith.science/pith/M4GQN4NWJTMQ3QWP6SZGOTHWZC/action/author_attestation","sign_citation":"https://pith.science/pith/M4GQN4NWJTMQ3QWP6SZGOTHWZC/action/citation_signature","submit_replication":"https://pith.science/pith/M4GQN4NWJTMQ3QWP6SZGOTHWZC/action/replication_record"}},"created_at":"2026-07-10T01:19:42.041145+00:00","updated_at":"2026-07-10T01:19:42.041145+00:00"}