{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2016:VK447RYRXSHZQQB2ZMLAMWYBXQ","short_pith_number":"pith:VK447RYR","schema_version":"1.0","canonical_sha256":"aab9cfc711bc8f98403acb16065b01bc07f83ceafa18336aca1c09dcd7c7064f","source":{"kind":"arxiv","id":"1611.01232","version":2},"attestation_state":"computed","paper":{"title":"Deep Information Propagation","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"stat.ML","authors_text":"Jascha Sohl-Dickstein, Justin Gilmer, Samuel S. Schoenholz, Surya Ganguli","submitted_at":"2016-11-04T00:44:32Z","abstract_excerpt":"We study the behavior of untrained neural networks whose weights and biases are randomly distributed using mean field theory. We show the existence of depth scales that naturally limit the maximum depth of signal propagation through these random networks. Our main practical result is to show that random networks may be trained precisely when information can travel through them. Thus, the depth scales that we identify provide bounds on how deep a network may be trained for a specific choice of hyperparameters. As a corollary to this, we argue that in networks at the edge of chaos, one of these "},"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":"1611.01232","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"stat.ML","submitted_at":"2016-11-04T00:44:32Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"4d24e98150282893c06be5698083a00bf9514b6db841534b10546bae6ad14f87","abstract_canon_sha256":"3697df88c0d003adc4fb1c6e70cb43ee5b70b3911eca593c1a84af92b30f2d23"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-18T00:47:00.338831Z","signature_b64":"+CFemjtSDnw2qGJ+jMYQW1+SaN6MRonsvI7kd8hlL1j2WBzowlRbRHAAdY7ItJGYpiILf0Q0bPXb/ayPEwdkAg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"aab9cfc711bc8f98403acb16065b01bc07f83ceafa18336aca1c09dcd7c7064f","last_reissued_at":"2026-05-18T00:47:00.337974Z","signature_status":"signed_v1","first_computed_at":"2026-05-18T00:47:00.337974Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Deep Information Propagation","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"stat.ML","authors_text":"Jascha Sohl-Dickstein, Justin Gilmer, Samuel S. Schoenholz, Surya Ganguli","submitted_at":"2016-11-04T00:44:32Z","abstract_excerpt":"We study the behavior of untrained neural networks whose weights and biases are randomly distributed using mean field theory. We show the existence of depth scales that naturally limit the maximum depth of signal propagation through these random networks. Our main practical result is to show that random networks may be trained precisely when information can travel through them. Thus, the depth scales that we identify provide bounds on how deep a network may be trained for a specific choice of hyperparameters. As a corollary to this, we argue that in networks at the edge of chaos, one of these "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"1611.01232","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":""},"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":"1611.01232","created_at":"2026-05-18T00:47:00.338124+00:00"},{"alias_kind":"arxiv_version","alias_value":"1611.01232v2","created_at":"2026-05-18T00:47:00.338124+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1611.01232","created_at":"2026-05-18T00:47:00.338124+00:00"},{"alias_kind":"pith_short_12","alias_value":"VK447RYRXSHZ","created_at":"2026-05-18T12:30:48.956258+00:00"},{"alias_kind":"pith_short_16","alias_value":"VK447RYRXSHZQQB2","created_at":"2026-05-18T12:30:48.956258+00:00"},{"alias_kind":"pith_short_8","alias_value":"VK447RYR","created_at":"2026-05-18T12:30:48.956258+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":5,"internal_anchor_count":3,"sample":[{"citing_arxiv_id":"2505.24333","citing_title":"Two failure modes of deep transformers and how to avoid them: a unified theory of signal propagation at initialisation","ref_index":1,"is_internal_anchor":true},{"citing_arxiv_id":"2511.06443","citing_title":"How Wide and How Deep? Mitigating Over-Squashing of GNNs via Channel Capacity Constrained Estimation","ref_index":45,"is_internal_anchor":true},{"citing_arxiv_id":"2605.14200","citing_title":"How to Scale Mixture-of-Experts: From muP to the Maximally Scale-Stable Parameterization","ref_index":101,"is_internal_anchor":true},{"citing_arxiv_id":"2605.05113","citing_title":"How Long Does Infinite Width Last? Signal Propagation in Long-Range Linear Recurrences","ref_index":33,"is_internal_anchor":false},{"citing_arxiv_id":"2604.11890","citing_title":"Subcritical Signal Propagation at Initialization in Normalization-Free Transformers","ref_index":3,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/VK447RYRXSHZQQB2ZMLAMWYBXQ","json":"https://pith.science/pith/VK447RYRXSHZQQB2ZMLAMWYBXQ.json","graph_json":"https://pith.science/api/pith-number/VK447RYRXSHZQQB2ZMLAMWYBXQ/graph.json","events_json":"https://pith.science/api/pith-number/VK447RYRXSHZQQB2ZMLAMWYBXQ/events.json","paper":"https://pith.science/paper/VK447RYR"},"agent_actions":{"view_html":"https://pith.science/pith/VK447RYRXSHZQQB2ZMLAMWYBXQ","download_json":"https://pith.science/pith/VK447RYRXSHZQQB2ZMLAMWYBXQ.json","view_paper":"https://pith.science/paper/VK447RYR","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=1611.01232&json=true","fetch_graph":"https://pith.science/api/pith-number/VK447RYRXSHZQQB2ZMLAMWYBXQ/graph.json","fetch_events":"https://pith.science/api/pith-number/VK447RYRXSHZQQB2ZMLAMWYBXQ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/VK447RYRXSHZQQB2ZMLAMWYBXQ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/VK447RYRXSHZQQB2ZMLAMWYBXQ/action/storage_attestation","attest_author":"https://pith.science/pith/VK447RYRXSHZQQB2ZMLAMWYBXQ/action/author_attestation","sign_citation":"https://pith.science/pith/VK447RYRXSHZQQB2ZMLAMWYBXQ/action/citation_signature","submit_replication":"https://pith.science/pith/VK447RYRXSHZQQB2ZMLAMWYBXQ/action/replication_record"}},"created_at":"2026-05-18T00:47:00.338124+00:00","updated_at":"2026-05-18T00:47:00.338124+00:00"}