{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:UEFUG46NTTKTSXOXTQWXEEP43J","short_pith_number":"pith:UEFUG46N","schema_version":"1.0","canonical_sha256":"a10b4373cd9cd5395dd79c2d7211fcda404b15e31a7112177c918e2ba90f0a0f","source":{"kind":"arxiv","id":"2602.11399","version":2},"attestation_state":"computed","paper":{"title":"Can We Really Learn One Representation to Optimize All Rewards?","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.RO"],"primary_cat":"cs.LG","authors_text":"Benjamin Eysenbach, Chongyi Zheng, Royina Karegoudra Jayanth","submitted_at":"2026-02-11T22:06:25Z","abstract_excerpt":"As unsupervised pretraining becomes increasingly ubiquitous in reinforcement learning, a more thorough theoretical understanding of these methods becomes of equal importance to their empirical success. We focus on the setting of unsupervised learning via interaction, where the forward-backward (FB) representation learning serves as a prototypical and popular example. In this paper, we shed light on FB by formally contextualizing the method within a broader class of recent methods that use regression to obtain a low-rank approximation of a successor measure ratio. Our analysis clarifies when FB"},"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":"2602.11399","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-02-11T22:06:25Z","cross_cats_sorted":["cs.AI","cs.RO"],"title_canon_sha256":"07ad9f4463ea38e490a7fb2fb6975b5debfeec0b4ffd3fdb90f01e920e6ce100","abstract_canon_sha256":"9bc4aa4c69b94ebf6e810e76e666abdf9fbe7f1276ca28bf747b91017b29ea33"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-09T00:19:36.685097Z","signature_b64":"cwit6slAHZlZT5B90c8MilCDQUJZTbDSQ3n7YNI3SenBJmC8HzF1+W8AkQ27rYUfOfIWOU6fiqGmxDEaMN4aAg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"a10b4373cd9cd5395dd79c2d7211fcda404b15e31a7112177c918e2ba90f0a0f","last_reissued_at":"2026-07-09T00:19:36.684600Z","signature_status":"signed_v1","first_computed_at":"2026-07-09T00:19:36.684600Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Can We Really Learn One Representation to Optimize All Rewards?","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.RO"],"primary_cat":"cs.LG","authors_text":"Benjamin Eysenbach, Chongyi Zheng, Royina Karegoudra Jayanth","submitted_at":"2026-02-11T22:06:25Z","abstract_excerpt":"As unsupervised pretraining becomes increasingly ubiquitous in reinforcement learning, a more thorough theoretical understanding of these methods becomes of equal importance to their empirical success. We focus on the setting of unsupervised learning via interaction, where the forward-backward (FB) representation learning serves as a prototypical and popular example. In this paper, we shed light on FB by formally contextualizing the method within a broader class of recent methods that use regression to obtain a low-rank approximation of a successor measure ratio. Our analysis clarifies when FB"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2602.11399","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/2602.11399/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":"2602.11399","created_at":"2026-07-09T00:19:36.684661+00:00"},{"alias_kind":"arxiv_version","alias_value":"2602.11399v2","created_at":"2026-07-09T00:19:36.684661+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2602.11399","created_at":"2026-07-09T00:19:36.684661+00:00"},{"alias_kind":"pith_short_12","alias_value":"UEFUG46NTTKT","created_at":"2026-07-09T00:19:36.684661+00:00"},{"alias_kind":"pith_short_16","alias_value":"UEFUG46NTTKTSXOX","created_at":"2026-07-09T00:19:36.684661+00:00"},{"alias_kind":"pith_short_8","alias_value":"UEFUG46N","created_at":"2026-07-09T00:19:36.684661+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":2,"sample":[{"citing_arxiv_id":"2606.11525","citing_title":"Learning Object Manipulation from Scratch via Contrastive Interaction","ref_index":65,"is_internal_anchor":true},{"citing_arxiv_id":"2605.13207","citing_title":"Switching Successor Measures for Hierarchical Zero-shot Reinforcement Learning","ref_index":62,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/UEFUG46NTTKTSXOXTQWXEEP43J","json":"https://pith.science/pith/UEFUG46NTTKTSXOXTQWXEEP43J.json","graph_json":"https://pith.science/api/pith-number/UEFUG46NTTKTSXOXTQWXEEP43J/graph.json","events_json":"https://pith.science/api/pith-number/UEFUG46NTTKTSXOXTQWXEEP43J/events.json","paper":"https://pith.science/paper/UEFUG46N"},"agent_actions":{"view_html":"https://pith.science/pith/UEFUG46NTTKTSXOXTQWXEEP43J","download_json":"https://pith.science/pith/UEFUG46NTTKTSXOXTQWXEEP43J.json","view_paper":"https://pith.science/paper/UEFUG46N","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2602.11399&json=true","fetch_graph":"https://pith.science/api/pith-number/UEFUG46NTTKTSXOXTQWXEEP43J/graph.json","fetch_events":"https://pith.science/api/pith-number/UEFUG46NTTKTSXOXTQWXEEP43J/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/UEFUG46NTTKTSXOXTQWXEEP43J/action/timestamp_anchor","attest_storage":"https://pith.science/pith/UEFUG46NTTKTSXOXTQWXEEP43J/action/storage_attestation","attest_author":"https://pith.science/pith/UEFUG46NTTKTSXOXTQWXEEP43J/action/author_attestation","sign_citation":"https://pith.science/pith/UEFUG46NTTKTSXOXTQWXEEP43J/action/citation_signature","submit_replication":"https://pith.science/pith/UEFUG46NTTKTSXOXTQWXEEP43J/action/replication_record"}},"created_at":"2026-07-09T00:19:36.684661+00:00","updated_at":"2026-07-09T00:19:36.684661+00:00"}