{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2021:SS5AHLJ5XKP6QNSCGRLB4GJMFP","short_pith_number":"pith:SS5AHLJ5","schema_version":"1.0","canonical_sha256":"94ba03ad3dba9fe8364234561e192c2bc4f08ca6765fc3cfedae0334e73e02e0","source":{"kind":"arxiv","id":"2102.12425","version":1},"attestation_state":"computed","paper":{"title":"Synthetic Returns for Long-Term Credit Assignment","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Adam Santoro, David Raposo, Francis Song, Greg Wayne, Hado van Hasselt, Matt Botvinick, Sam Ritter, Theophane Weber","submitted_at":"2021-02-24T17:43:02Z","abstract_excerpt":"Since the earliest days of reinforcement learning, the workhorse method for assigning credit to actions over time has been temporal-difference (TD) learning, which propagates credit backward timestep-by-timestep. This approach suffers when delays between actions and rewards are long and when intervening unrelated events contribute variance to long-term returns. We propose state-associative (SA) learning, where the agent learns associations between states and arbitrarily distant future rewards, then propagates credit directly between the two. In this work, we use SA-learning to model the contri"},"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":"2102.12425","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2021-02-24T17:43:02Z","cross_cats_sorted":[],"title_canon_sha256":"bdba32908aef5d83de0d426535783f0635f77f8628a05a91e6dd5da79a213721","abstract_canon_sha256":"e8323e68dbc00840d39525968d25ee1b5f3bd10850b9b0170688a9da940d47a7"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T02:18:12.686924Z","signature_b64":"f1PnbmkYbGM5Hk9fEYvL5iPF2qPgaUJo77dWU7dVU7yrPXV3yXPiPy8gTUp+NKO29+0awt8qOt0+QQHU47XjDQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"94ba03ad3dba9fe8364234561e192c2bc4f08ca6765fc3cfedae0334e73e02e0","last_reissued_at":"2026-07-05T02:18:12.686449Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T02:18:12.686449Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Synthetic Returns for Long-Term Credit Assignment","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Adam Santoro, David Raposo, Francis Song, Greg Wayne, Hado van Hasselt, Matt Botvinick, Sam Ritter, Theophane Weber","submitted_at":"2021-02-24T17:43:02Z","abstract_excerpt":"Since the earliest days of reinforcement learning, the workhorse method for assigning credit to actions over time has been temporal-difference (TD) learning, which propagates credit backward timestep-by-timestep. This approach suffers when delays between actions and rewards are long and when intervening unrelated events contribute variance to long-term returns. We propose state-associative (SA) learning, where the agent learns associations between states and arbitrarily distant future rewards, then propagates credit directly between the two. In this work, we use SA-learning to model the contri"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2102.12425","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/2102.12425/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":"2102.12425","created_at":"2026-07-05T02:18:12.686511+00:00"},{"alias_kind":"arxiv_version","alias_value":"2102.12425v1","created_at":"2026-07-05T02:18:12.686511+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2102.12425","created_at":"2026-07-05T02:18:12.686511+00:00"},{"alias_kind":"pith_short_12","alias_value":"SS5AHLJ5XKP6","created_at":"2026-07-05T02:18:12.686511+00:00"},{"alias_kind":"pith_short_16","alias_value":"SS5AHLJ5XKP6QNSC","created_at":"2026-07-05T02:18:12.686511+00:00"},{"alias_kind":"pith_short_8","alias_value":"SS5AHLJ5","created_at":"2026-07-05T02:18:12.686511+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2106.01345","citing_title":"Decision Transformer: Reinforcement Learning via Sequence Modeling","ref_index":25,"is_internal_anchor":false},{"citing_arxiv_id":"2604.04983","citing_title":"Territory Paint Wars: Diagnosing and Mitigating Failure Modes in Competitive Multi-Agent PPO","ref_index":3,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/SS5AHLJ5XKP6QNSCGRLB4GJMFP","json":"https://pith.science/pith/SS5AHLJ5XKP6QNSCGRLB4GJMFP.json","graph_json":"https://pith.science/api/pith-number/SS5AHLJ5XKP6QNSCGRLB4GJMFP/graph.json","events_json":"https://pith.science/api/pith-number/SS5AHLJ5XKP6QNSCGRLB4GJMFP/events.json","paper":"https://pith.science/paper/SS5AHLJ5"},"agent_actions":{"view_html":"https://pith.science/pith/SS5AHLJ5XKP6QNSCGRLB4GJMFP","download_json":"https://pith.science/pith/SS5AHLJ5XKP6QNSCGRLB4GJMFP.json","view_paper":"https://pith.science/paper/SS5AHLJ5","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2102.12425&json=true","fetch_graph":"https://pith.science/api/pith-number/SS5AHLJ5XKP6QNSCGRLB4GJMFP/graph.json","fetch_events":"https://pith.science/api/pith-number/SS5AHLJ5XKP6QNSCGRLB4GJMFP/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/SS5AHLJ5XKP6QNSCGRLB4GJMFP/action/timestamp_anchor","attest_storage":"https://pith.science/pith/SS5AHLJ5XKP6QNSCGRLB4GJMFP/action/storage_attestation","attest_author":"https://pith.science/pith/SS5AHLJ5XKP6QNSCGRLB4GJMFP/action/author_attestation","sign_citation":"https://pith.science/pith/SS5AHLJ5XKP6QNSCGRLB4GJMFP/action/citation_signature","submit_replication":"https://pith.science/pith/SS5AHLJ5XKP6QNSCGRLB4GJMFP/action/replication_record"}},"created_at":"2026-07-05T02:18:12.686511+00:00","updated_at":"2026-07-05T02:18:12.686511+00:00"}