{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2020:44XSPQVVSLAPYVOJPOUPQSVVJ7","short_pith_number":"pith:44XSPQVV","schema_version":"1.0","canonical_sha256":"e72f27c2b592c0fc55c97ba8f84ab54fd76fd4f7481bc2c41f5cb8c7012f94a6","source":{"kind":"arxiv","id":"2006.14548","version":4},"attestation_state":"computed","paper":{"title":"Tensor Programs II: Neural Tangent Kernel for Any Architecture","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cond-mat.dis-nn","cs.LG","cs.NE"],"primary_cat":"stat.ML","authors_text":"Greg Yang","submitted_at":"2020-06-25T16:45:23Z","abstract_excerpt":"We prove that a randomly initialized neural network of *any architecture* has its Tangent Kernel (NTK) converge to a deterministic limit, as the network widths tend to infinity. We demonstrate how to calculate this limit. In prior literature, the heuristic study of neural network gradients often assumes every weight matrix used in forward propagation is independent from its transpose used in backpropagation (Schoenholz et al. 2017). This is known as the *gradient independence assumption (GIA)*. We identify a commonly satisfied condition, which we call *Simple GIA Check*, such that the NTK limi"},"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":"2006.14548","kind":"arxiv","version":4},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"stat.ML","submitted_at":"2020-06-25T16:45:23Z","cross_cats_sorted":["cond-mat.dis-nn","cs.LG","cs.NE"],"title_canon_sha256":"aee3b85e2917517e2ddd511345ec688ace8556d49545a365084e6298ef81d697","abstract_canon_sha256":"0ee52a40c9fb94a423c1a55d943709f234fb488d4a04cd2a88dd007fc62f1b78"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T01:55:10.962620Z","signature_b64":"b39x4pnE/MwsA7KJ3pvnVNwHnjd5C64z8vzc8zdgPTwpY98GmwL1grlFUlWOukQyCqRBkN+JBQgv6yA799olAA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"e72f27c2b592c0fc55c97ba8f84ab54fd76fd4f7481bc2c41f5cb8c7012f94a6","last_reissued_at":"2026-07-05T01:55:10.962240Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T01:55:10.962240Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Tensor Programs II: Neural Tangent Kernel for Any Architecture","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cond-mat.dis-nn","cs.LG","cs.NE"],"primary_cat":"stat.ML","authors_text":"Greg Yang","submitted_at":"2020-06-25T16:45:23Z","abstract_excerpt":"We prove that a randomly initialized neural network of *any architecture* has its Tangent Kernel (NTK) converge to a deterministic limit, as the network widths tend to infinity. We demonstrate how to calculate this limit. In prior literature, the heuristic study of neural network gradients often assumes every weight matrix used in forward propagation is independent from its transpose used in backpropagation (Schoenholz et al. 2017). This is known as the *gradient independence assumption (GIA)*. We identify a commonly satisfied condition, which we call *Simple GIA Check*, such that the NTK limi"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2006.14548","kind":"arxiv","version":4},"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/2006.14548/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":"2006.14548","created_at":"2026-07-05T01:55:10.962295+00:00"},{"alias_kind":"arxiv_version","alias_value":"2006.14548v4","created_at":"2026-07-05T01:55:10.962295+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2006.14548","created_at":"2026-07-05T01:55:10.962295+00:00"},{"alias_kind":"pith_short_12","alias_value":"44XSPQVVSLAP","created_at":"2026-07-05T01:55:10.962295+00:00"},{"alias_kind":"pith_short_16","alias_value":"44XSPQVVSLAPYVOJ","created_at":"2026-07-05T01:55:10.962295+00:00"},{"alias_kind":"pith_short_8","alias_value":"44XSPQVV","created_at":"2026-07-05T01:55:10.962295+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":11,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.06418","citing_title":"Double Preconditioning (DoPr): Optimization for Test-Time Performance, not Validation Loss","ref_index":169,"is_internal_anchor":false},{"citing_arxiv_id":"2605.17606","citing_title":"The Neural Tangent Kernel for Classification","ref_index":6,"is_internal_anchor":false},{"citing_arxiv_id":"2508.03810","citing_title":"Viability of perturbative expansion for quantum field theories on neurons","ref_index":31,"is_internal_anchor":false},{"citing_arxiv_id":"2605.17606","citing_title":"The Neural Tangent Kernel for Classification","ref_index":6,"is_internal_anchor":false},{"citing_arxiv_id":"2605.18180","citing_title":"Canonical Regularisation of Wide Feature-Learning Neural Networks","ref_index":47,"is_internal_anchor":false},{"citing_arxiv_id":"2605.15290","citing_title":"GQA-{\\mu}P: The maximal parameterization update for grouped query attention","ref_index":18,"is_internal_anchor":false},{"citing_arxiv_id":"2603.28964","citing_title":"Spectral Edge Dynamics: An Analytical-Empirical Study of Phase Transitions in Neural Network Training","ref_index":22,"is_internal_anchor":false},{"citing_arxiv_id":"2605.12763","citing_title":"State-Space NTK Collapse Near Bifurcations","ref_index":123,"is_internal_anchor":false},{"citing_arxiv_id":"2604.27077","citing_title":"Learning Rate Transfer in Normalized Transformers","ref_index":19,"is_internal_anchor":false},{"citing_arxiv_id":"2605.08746","citing_title":"The Global Empirical NTK: Self-Referential Bias and Dimensionality of Gradient Descent Learning","ref_index":124,"is_internal_anchor":false},{"citing_arxiv_id":"2605.05113","citing_title":"How Long Does Infinite Width Last? Signal Propagation in Long-Range Linear Recurrences","ref_index":42,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/44XSPQVVSLAPYVOJPOUPQSVVJ7","json":"https://pith.science/pith/44XSPQVVSLAPYVOJPOUPQSVVJ7.json","graph_json":"https://pith.science/api/pith-number/44XSPQVVSLAPYVOJPOUPQSVVJ7/graph.json","events_json":"https://pith.science/api/pith-number/44XSPQVVSLAPYVOJPOUPQSVVJ7/events.json","paper":"https://pith.science/paper/44XSPQVV"},"agent_actions":{"view_html":"https://pith.science/pith/44XSPQVVSLAPYVOJPOUPQSVVJ7","download_json":"https://pith.science/pith/44XSPQVVSLAPYVOJPOUPQSVVJ7.json","view_paper":"https://pith.science/paper/44XSPQVV","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2006.14548&json=true","fetch_graph":"https://pith.science/api/pith-number/44XSPQVVSLAPYVOJPOUPQSVVJ7/graph.json","fetch_events":"https://pith.science/api/pith-number/44XSPQVVSLAPYVOJPOUPQSVVJ7/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/44XSPQVVSLAPYVOJPOUPQSVVJ7/action/timestamp_anchor","attest_storage":"https://pith.science/pith/44XSPQVVSLAPYVOJPOUPQSVVJ7/action/storage_attestation","attest_author":"https://pith.science/pith/44XSPQVVSLAPYVOJPOUPQSVVJ7/action/author_attestation","sign_citation":"https://pith.science/pith/44XSPQVVSLAPYVOJPOUPQSVVJ7/action/citation_signature","submit_replication":"https://pith.science/pith/44XSPQVVSLAPYVOJPOUPQSVVJ7/action/replication_record"}},"created_at":"2026-07-05T01:55:10.962295+00:00","updated_at":"2026-07-05T01:55:10.962295+00:00"}