{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2020:F3ZUAXMOW26FD6TCV6NXRVZOII","short_pith_number":"pith:F3ZUAXMO","schema_version":"1.0","canonical_sha256":"2ef3405d8eb6bc51fa62af9b78d72e423be865856d2785d38f4ba4c0974da30f","source":{"kind":"arxiv","id":"2003.04887","version":2},"attestation_state":"computed","paper":{"title":"ReZero is All You Need: Fast Convergence at Large Depth","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CL","stat.ML"],"primary_cat":"cs.LG","authors_text":"Bodhisattwa Prasad Majumder, Garrison W. Cottrell, Huanru Henry Mao, Julian McAuley, Thomas Bachlechner","submitted_at":"2020-03-10T17:58:01Z","abstract_excerpt":"Deep networks often suffer from vanishing or exploding gradients due to inefficient signal propagation, leading to long training times or convergence difficulties. Various architecture designs, sophisticated residual-style networks, and initialization schemes have been shown to improve deep signal propagation. Recently, Pennington et al. used free probability theory to show that dynamical isometry plays an integral role in efficient deep learning. We show that the simplest architecture change of gating each residual connection using a single zero-initialized parameter satisfies initial dynamic"},"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":"2003.04887","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2020-03-10T17:58:01Z","cross_cats_sorted":["cs.CL","stat.ML"],"title_canon_sha256":"59e49d16178000e8908e7710d555df4004c4d272757049d59e1b80bc3a391e59","abstract_canon_sha256":"d4711a14ea4160241e853b03a5f0bbae1bb470ac42f7a9ea211455089625377f"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T01:13:25.472049Z","signature_b64":"VVS1Dfjv9lNSoYPJnoUUJJLbY4E05svoGiJyuhK6JKTKWLkZwQeajtGEjCltd1HdnwRw8JjtrYP9mfg9BOg0BA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"2ef3405d8eb6bc51fa62af9b78d72e423be865856d2785d38f4ba4c0974da30f","last_reissued_at":"2026-07-05T01:13:25.471600Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T01:13:25.471600Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"ReZero is All You Need: Fast Convergence at Large Depth","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CL","stat.ML"],"primary_cat":"cs.LG","authors_text":"Bodhisattwa Prasad Majumder, Garrison W. Cottrell, Huanru Henry Mao, Julian McAuley, Thomas Bachlechner","submitted_at":"2020-03-10T17:58:01Z","abstract_excerpt":"Deep networks often suffer from vanishing or exploding gradients due to inefficient signal propagation, leading to long training times or convergence difficulties. Various architecture designs, sophisticated residual-style networks, and initialization schemes have been shown to improve deep signal propagation. Recently, Pennington et al. used free probability theory to show that dynamical isometry plays an integral role in efficient deep learning. We show that the simplest architecture change of gating each residual connection using a single zero-initialized parameter satisfies initial dynamic"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2003.04887","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/2003.04887/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":"2003.04887","created_at":"2026-07-05T01:13:25.471676+00:00"},{"alias_kind":"arxiv_version","alias_value":"2003.04887v2","created_at":"2026-07-05T01:13:25.471676+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2003.04887","created_at":"2026-07-05T01:13:25.471676+00:00"},{"alias_kind":"pith_short_12","alias_value":"F3ZUAXMOW26F","created_at":"2026-07-05T01:13:25.471676+00:00"},{"alias_kind":"pith_short_16","alias_value":"F3ZUAXMOW26FD6TC","created_at":"2026-07-05T01:13:25.471676+00:00"},{"alias_kind":"pith_short_8","alias_value":"F3ZUAXMO","created_at":"2026-07-05T01:13:25.471676+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":9,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.19907","citing_title":"NNNN: Neural Networks for Newtonian Noise Mitigation at the Einstein Telescope","ref_index":26,"is_internal_anchor":false},{"citing_arxiv_id":"2606.06564","citing_title":"HAARES Half-Split Residual Basis Routing for Deep Transformers","ref_index":4,"is_internal_anchor":false},{"citing_arxiv_id":"2606.31859","citing_title":"Review Residuals: Update-Conditioned Residual Gating for Transformers","ref_index":4,"is_internal_anchor":false},{"citing_arxiv_id":"2602.10408","citing_title":"Gated Normalization Removal and Scale Anchoring in Pre-Norm Transformers","ref_index":1,"is_internal_anchor":false},{"citing_arxiv_id":"2603.15031","citing_title":"Attention Residuals","ref_index":2,"is_internal_anchor":false},{"citing_arxiv_id":"2603.20120","citing_title":"Deep learning-based phase-field modelling of brittle fracture in anisotropic media","ref_index":33,"is_internal_anchor":false},{"citing_arxiv_id":"2605.06047","citing_title":"TFM-Retouche: A Lightweight Input-Space Adapter for Tabular Foundation Models","ref_index":2,"is_internal_anchor":false},{"citing_arxiv_id":"2605.06047","citing_title":"TFM-Retouche: A Lightweight Input-Space Adapter for Tabular Foundation Models","ref_index":1,"is_internal_anchor":false},{"citing_arxiv_id":"2604.22117","citing_title":"PermaFrost-Attack: Stealth Pretraining Seeding(SPS) for planting Logic Landmines During LLM Training","ref_index":19,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/F3ZUAXMOW26FD6TCV6NXRVZOII","json":"https://pith.science/pith/F3ZUAXMOW26FD6TCV6NXRVZOII.json","graph_json":"https://pith.science/api/pith-number/F3ZUAXMOW26FD6TCV6NXRVZOII/graph.json","events_json":"https://pith.science/api/pith-number/F3ZUAXMOW26FD6TCV6NXRVZOII/events.json","paper":"https://pith.science/paper/F3ZUAXMO"},"agent_actions":{"view_html":"https://pith.science/pith/F3ZUAXMOW26FD6TCV6NXRVZOII","download_json":"https://pith.science/pith/F3ZUAXMOW26FD6TCV6NXRVZOII.json","view_paper":"https://pith.science/paper/F3ZUAXMO","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2003.04887&json=true","fetch_graph":"https://pith.science/api/pith-number/F3ZUAXMOW26FD6TCV6NXRVZOII/graph.json","fetch_events":"https://pith.science/api/pith-number/F3ZUAXMOW26FD6TCV6NXRVZOII/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/F3ZUAXMOW26FD6TCV6NXRVZOII/action/timestamp_anchor","attest_storage":"https://pith.science/pith/F3ZUAXMOW26FD6TCV6NXRVZOII/action/storage_attestation","attest_author":"https://pith.science/pith/F3ZUAXMOW26FD6TCV6NXRVZOII/action/author_attestation","sign_citation":"https://pith.science/pith/F3ZUAXMOW26FD6TCV6NXRVZOII/action/citation_signature","submit_replication":"https://pith.science/pith/F3ZUAXMOW26FD6TCV6NXRVZOII/action/replication_record"}},"created_at":"2026-07-05T01:13:25.471676+00:00","updated_at":"2026-07-05T01:13:25.471676+00:00"}