{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2020:JKYM5R5NTL4U3OR2N3FXDTNDJF","short_pith_number":"pith:JKYM5R5N","schema_version":"1.0","canonical_sha256":"4ab0cec7ad9af94dba3a6ecb71cda3497ece66083af501e5b4c01e4d6d3f1035","source":{"kind":"arxiv","id":"2009.11243","version":1},"attestation_state":"computed","paper":{"title":"Tasks, stability, architecture, and compute: Training more effective learned optimizers, and using them to train themselves","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.NE","stat.ML"],"primary_cat":"cs.LG","authors_text":"Ben Poole, C. Daniel Freeman, Jascha Sohl-Dickstein, Luke Metz, Niru Maheswaranathan","submitted_at":"2020-09-23T16:35:09Z","abstract_excerpt":"Much as replacing hand-designed features with learned functions has revolutionized how we solve perceptual tasks, we believe learned algorithms will transform how we train models. In this work we focus on general-purpose learned optimizers capable of training a wide variety of problems with no user-specified hyperparameters. We introduce a new, neural network parameterized, hierarchical optimizer with access to additional features such as validation loss to enable automatic regularization. Most learned optimizers have been trained on only a single task, or a small number of tasks. We train our"},"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":"2009.11243","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2020-09-23T16:35:09Z","cross_cats_sorted":["cs.NE","stat.ML"],"title_canon_sha256":"23de0beb6115fcdde081400991c8a621c3dd072553c7831f2d3106a6ffd6f936","abstract_canon_sha256":"7f44d0b468fe17daaf14dd311626b7c39e74b66a355beb0fd794bfe284901f88"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T01:37:37.608217Z","signature_b64":"85iBGRC0OuSWdlx8+cXuKtWO4gqUIqM6h5rCB3/C8K6MqDaF0xdJ1hTy2BnHGbmUbzl1kZbulfFi3KTQvgUSBQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"4ab0cec7ad9af94dba3a6ecb71cda3497ece66083af501e5b4c01e4d6d3f1035","last_reissued_at":"2026-07-05T01:37:37.607813Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T01:37:37.607813Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Tasks, stability, architecture, and compute: Training more effective learned optimizers, and using them to train themselves","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.NE","stat.ML"],"primary_cat":"cs.LG","authors_text":"Ben Poole, C. Daniel Freeman, Jascha Sohl-Dickstein, Luke Metz, Niru Maheswaranathan","submitted_at":"2020-09-23T16:35:09Z","abstract_excerpt":"Much as replacing hand-designed features with learned functions has revolutionized how we solve perceptual tasks, we believe learned algorithms will transform how we train models. In this work we focus on general-purpose learned optimizers capable of training a wide variety of problems with no user-specified hyperparameters. We introduce a new, neural network parameterized, hierarchical optimizer with access to additional features such as validation loss to enable automatic regularization. Most learned optimizers have been trained on only a single task, or a small number of tasks. We train our"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2009.11243","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/2009.11243/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":"2009.11243","created_at":"2026-07-05T01:37:37.607868+00:00"},{"alias_kind":"arxiv_version","alias_value":"2009.11243v1","created_at":"2026-07-05T01:37:37.607868+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2009.11243","created_at":"2026-07-05T01:37:37.607868+00:00"},{"alias_kind":"pith_short_12","alias_value":"JKYM5R5NTL4U","created_at":"2026-07-05T01:37:37.607868+00:00"},{"alias_kind":"pith_short_16","alias_value":"JKYM5R5NTL4U3OR2","created_at":"2026-07-05T01:37:37.607868+00:00"},{"alias_kind":"pith_short_8","alias_value":"JKYM5R5N","created_at":"2026-07-05T01:37:37.607868+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.20404","citing_title":"FlowBender: Feedback-Aware Training for Self-Correcting Conditional Flows","ref_index":41,"is_internal_anchor":false},{"citing_arxiv_id":"2605.15760","citing_title":"Learn2Splat: Extending the Horizon of Learned 3DGS Optimization","ref_index":34,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/JKYM5R5NTL4U3OR2N3FXDTNDJF","json":"https://pith.science/pith/JKYM5R5NTL4U3OR2N3FXDTNDJF.json","graph_json":"https://pith.science/api/pith-number/JKYM5R5NTL4U3OR2N3FXDTNDJF/graph.json","events_json":"https://pith.science/api/pith-number/JKYM5R5NTL4U3OR2N3FXDTNDJF/events.json","paper":"https://pith.science/paper/JKYM5R5N"},"agent_actions":{"view_html":"https://pith.science/pith/JKYM5R5NTL4U3OR2N3FXDTNDJF","download_json":"https://pith.science/pith/JKYM5R5NTL4U3OR2N3FXDTNDJF.json","view_paper":"https://pith.science/paper/JKYM5R5N","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2009.11243&json=true","fetch_graph":"https://pith.science/api/pith-number/JKYM5R5NTL4U3OR2N3FXDTNDJF/graph.json","fetch_events":"https://pith.science/api/pith-number/JKYM5R5NTL4U3OR2N3FXDTNDJF/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/JKYM5R5NTL4U3OR2N3FXDTNDJF/action/timestamp_anchor","attest_storage":"https://pith.science/pith/JKYM5R5NTL4U3OR2N3FXDTNDJF/action/storage_attestation","attest_author":"https://pith.science/pith/JKYM5R5NTL4U3OR2N3FXDTNDJF/action/author_attestation","sign_citation":"https://pith.science/pith/JKYM5R5NTL4U3OR2N3FXDTNDJF/action/citation_signature","submit_replication":"https://pith.science/pith/JKYM5R5NTL4U3OR2N3FXDTNDJF/action/replication_record"}},"created_at":"2026-07-05T01:37:37.607868+00:00","updated_at":"2026-07-05T01:37:37.607868+00:00"}