{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2015:INEJ364T7UXLUUHT6XGTGX77PX","short_pith_number":"pith:INEJ364T","schema_version":"1.0","canonical_sha256":"43489dfb93fd2eba50f3f5cd335fff7df5a8f1081e4500765df528c016177ca8","source":{"kind":"arxiv","id":"1502.03492","version":3},"attestation_state":"computed","paper":{"title":"Gradient-based Hyperparameter Optimization through Reversible Learning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"stat.ML","authors_text":"David Duvenaud, Dougal Maclaurin, Ryan P. Adams","submitted_at":"2015-02-11T23:52:36Z","abstract_excerpt":"Tuning hyperparameters of learning algorithms is hard because gradients are usually unavailable. We compute exact gradients of cross-validation performance with respect to all hyperparameters by chaining derivatives backwards through the entire training procedure. These gradients allow us to optimize thousands of hyperparameters, including step-size and momentum schedules, weight initialization distributions, richly parameterized regularization schemes, and neural network architectures. We compute hyperparameter gradients by exactly reversing the dynamics of stochastic gradient descent with mo"},"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":"1502.03492","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"stat.ML","submitted_at":"2015-02-11T23:52:36Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"40c11503365b84caa245e8e4795cadd15c92cbe65e15389ef3a5aca7a5ea7846","abstract_canon_sha256":"b1510a829bff56b93789048df089b00d5cbb05f34c0f1eb5cdc39f08301972d9"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-18T02:19:43.575219Z","signature_b64":"cOF9PmfezF0xl0krzqwwOXHqWl8q3CWuerdcsTBxbWfBasiqBw3if4sEjREW9xgsKq+fyaMKu0apwe/8toReCg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"43489dfb93fd2eba50f3f5cd335fff7df5a8f1081e4500765df528c016177ca8","last_reissued_at":"2026-05-18T02:19:43.574789Z","signature_status":"signed_v1","first_computed_at":"2026-05-18T02:19:43.574789Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Gradient-based Hyperparameter Optimization through Reversible Learning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"stat.ML","authors_text":"David Duvenaud, Dougal Maclaurin, Ryan P. Adams","submitted_at":"2015-02-11T23:52:36Z","abstract_excerpt":"Tuning hyperparameters of learning algorithms is hard because gradients are usually unavailable. We compute exact gradients of cross-validation performance with respect to all hyperparameters by chaining derivatives backwards through the entire training procedure. These gradients allow us to optimize thousands of hyperparameters, including step-size and momentum schedules, weight initialization distributions, richly parameterized regularization schemes, and neural network architectures. We compute hyperparameter gradients by exactly reversing the dynamics of stochastic gradient descent with mo"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"1502.03492","kind":"arxiv","version":3},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"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":"1502.03492","created_at":"2026-05-18T02:19:43.574845+00:00"},{"alias_kind":"arxiv_version","alias_value":"1502.03492v3","created_at":"2026-05-18T02:19:43.574845+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1502.03492","created_at":"2026-05-18T02:19:43.574845+00:00"},{"alias_kind":"pith_short_12","alias_value":"INEJ364T7UXL","created_at":"2026-05-18T12:29:25.134429+00:00"},{"alias_kind":"pith_short_16","alias_value":"INEJ364T7UXLUUHT","created_at":"2026-05-18T12:29:25.134429+00:00"},{"alias_kind":"pith_short_8","alias_value":"INEJ364T","created_at":"2026-05-18T12:29:25.134429+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2605.19195","citing_title":"The Thermodynamic Costs of Simple Linear Regression","ref_index":36,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/INEJ364T7UXLUUHT6XGTGX77PX","json":"https://pith.science/pith/INEJ364T7UXLUUHT6XGTGX77PX.json","graph_json":"https://pith.science/api/pith-number/INEJ364T7UXLUUHT6XGTGX77PX/graph.json","events_json":"https://pith.science/api/pith-number/INEJ364T7UXLUUHT6XGTGX77PX/events.json","paper":"https://pith.science/paper/INEJ364T"},"agent_actions":{"view_html":"https://pith.science/pith/INEJ364T7UXLUUHT6XGTGX77PX","download_json":"https://pith.science/pith/INEJ364T7UXLUUHT6XGTGX77PX.json","view_paper":"https://pith.science/paper/INEJ364T","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=1502.03492&json=true","fetch_graph":"https://pith.science/api/pith-number/INEJ364T7UXLUUHT6XGTGX77PX/graph.json","fetch_events":"https://pith.science/api/pith-number/INEJ364T7UXLUUHT6XGTGX77PX/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/INEJ364T7UXLUUHT6XGTGX77PX/action/timestamp_anchor","attest_storage":"https://pith.science/pith/INEJ364T7UXLUUHT6XGTGX77PX/action/storage_attestation","attest_author":"https://pith.science/pith/INEJ364T7UXLUUHT6XGTGX77PX/action/author_attestation","sign_citation":"https://pith.science/pith/INEJ364T7UXLUUHT6XGTGX77PX/action/citation_signature","submit_replication":"https://pith.science/pith/INEJ364T7UXLUUHT6XGTGX77PX/action/replication_record"}},"created_at":"2026-05-18T02:19:43.574845+00:00","updated_at":"2026-05-18T02:19:43.574845+00:00"}