{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:HS6JDM5DEZTKIKEKQ3LR5F24ZV","short_pith_number":"pith:HS6JDM5D","schema_version":"1.0","canonical_sha256":"3cbc91b3a32666a4288a86d71e975ccd51235141ee2c8e7e234915be5409be55","source":{"kind":"arxiv","id":"2607.16761","version":1},"attestation_state":"computed","paper":{"title":"Dropout and Random Gradient Masking Are Asymptotically Equivalent in Large ResNets","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG","math.PR"],"primary_cat":"stat.ML","authors_text":"Javier Maass, L\\'ena\\\"ic Chizat","submitted_at":"2026-07-18T10:57:27Z","abstract_excerpt":"Dropout and Random Gradient Masking (RaM) are two training techniques used to improve performance in deep learning. Both techniques inject randomness into the training dynamics, but in significantly different ways: dropout applies random masks to the activations in the forward pass, whereas RaM leaves the forward pass unchanged and instead masks the gradients. In particular, the noise induced by RaM in the parameter updates is unbiased, so standard explanations for the effectiveness of dropout, such as the penalization effect or the prevention of co-adaptation between neurons, do not apply to "},"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":"2607.16761","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"stat.ML","submitted_at":"2026-07-18T10:57:27Z","cross_cats_sorted":["cs.LG","math.PR"],"title_canon_sha256":"6dc85e6a8207f0d84979387861fe43c920afe33e4362c038a39560230ac0613e","abstract_canon_sha256":"f8bd602121adfcaac677c431af0f2a51544c5627449ea1c3b8a7603a7c425c68"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-21T01:20:57.745263Z","signature_b64":"hqGreRbdoG7ZuRf6RbgWx4HXOCl4HnJwYUYVvzWFkEFtXhQfukYY6aFCN6gmLaRQFHwOSUyfwPCuinIwl9RxBQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"3cbc91b3a32666a4288a86d71e975ccd51235141ee2c8e7e234915be5409be55","last_reissued_at":"2026-07-21T01:20:57.744310Z","signature_status":"signed_v1","first_computed_at":"2026-07-21T01:20:57.744310Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Dropout and Random Gradient Masking Are Asymptotically Equivalent in Large ResNets","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG","math.PR"],"primary_cat":"stat.ML","authors_text":"Javier Maass, L\\'ena\\\"ic Chizat","submitted_at":"2026-07-18T10:57:27Z","abstract_excerpt":"Dropout and Random Gradient Masking (RaM) are two training techniques used to improve performance in deep learning. Both techniques inject randomness into the training dynamics, but in significantly different ways: dropout applies random masks to the activations in the forward pass, whereas RaM leaves the forward pass unchanged and instead masks the gradients. In particular, the noise induced by RaM in the parameter updates is unbiased, so standard explanations for the effectiveness of dropout, such as the penalization effect or the prevention of co-adaptation between neurons, do not apply to "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2607.16761","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/2607.16761/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":"2607.16761","created_at":"2026-07-21T01:20:57.744758+00:00"},{"alias_kind":"arxiv_version","alias_value":"2607.16761v1","created_at":"2026-07-21T01:20:57.744758+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2607.16761","created_at":"2026-07-21T01:20:57.744758+00:00"},{"alias_kind":"pith_short_12","alias_value":"HS6JDM5DEZTK","created_at":"2026-07-21T01:20:57.744758+00:00"},{"alias_kind":"pith_short_16","alias_value":"HS6JDM5DEZTKIKEK","created_at":"2026-07-21T01:20:57.744758+00:00"},{"alias_kind":"pith_short_8","alias_value":"HS6JDM5D","created_at":"2026-07-21T01:20:57.744758+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/HS6JDM5DEZTKIKEKQ3LR5F24ZV","json":"https://pith.science/pith/HS6JDM5DEZTKIKEKQ3LR5F24ZV.json","graph_json":"https://pith.science/api/pith-number/HS6JDM5DEZTKIKEKQ3LR5F24ZV/graph.json","events_json":"https://pith.science/api/pith-number/HS6JDM5DEZTKIKEKQ3LR5F24ZV/events.json","paper":"https://pith.science/paper/HS6JDM5D"},"agent_actions":{"view_html":"https://pith.science/pith/HS6JDM5DEZTKIKEKQ3LR5F24ZV","download_json":"https://pith.science/pith/HS6JDM5DEZTKIKEKQ3LR5F24ZV.json","view_paper":"https://pith.science/paper/HS6JDM5D","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2607.16761&json=true","fetch_graph":"https://pith.science/api/pith-number/HS6JDM5DEZTKIKEKQ3LR5F24ZV/graph.json","fetch_events":"https://pith.science/api/pith-number/HS6JDM5DEZTKIKEKQ3LR5F24ZV/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/HS6JDM5DEZTKIKEKQ3LR5F24ZV/action/timestamp_anchor","attest_storage":"https://pith.science/pith/HS6JDM5DEZTKIKEKQ3LR5F24ZV/action/storage_attestation","attest_author":"https://pith.science/pith/HS6JDM5DEZTKIKEKQ3LR5F24ZV/action/author_attestation","sign_citation":"https://pith.science/pith/HS6JDM5DEZTKIKEKQ3LR5F24ZV/action/citation_signature","submit_replication":"https://pith.science/pith/HS6JDM5DEZTKIKEKQ3LR5F24ZV/action/replication_record"}},"created_at":"2026-07-21T01:20:57.744758+00:00","updated_at":"2026-07-21T01:20:57.744758+00:00"}