{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:UWKERGVLOISRV6ZCGJMUEX6L64","short_pith_number":"pith:UWKERGVL","schema_version":"1.0","canonical_sha256":"a594489aab72251afb223259425fcbf73f1397a0fa30b0fb625c18c086c3265f","source":{"kind":"arxiv","id":"2310.02244","version":5},"attestation_state":"computed","paper":{"title":"Tensor Programs VI: Feature Learning in Infinite-Depth Neural Networks","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cond-mat.dis-nn","math.PR"],"primary_cat":"cs.NE","authors_text":"Chen Zhu, Dingli Yu, Greg Yang, Soufiane Hayou","submitted_at":"2023-10-03T17:50:40Z","abstract_excerpt":"By classifying infinite-width neural networks and identifying the *optimal* limit, Tensor Programs IV and V demonstrated a universal way, called $\\mu$P, for *widthwise hyperparameter transfer*, i.e., predicting optimal hyperparameters of wide neural networks from narrow ones. Here we investigate the analogous classification for *depthwise parametrizations* of deep residual networks (resnets). We classify depthwise parametrizations of block multiplier and learning rate by their infinite-width-then-depth limits. In resnets where each block has only one layer, we identify a unique optimal paramet"},"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":"2310.02244","kind":"arxiv","version":5},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.NE","submitted_at":"2023-10-03T17:50:40Z","cross_cats_sorted":["cond-mat.dis-nn","math.PR"],"title_canon_sha256":"1bf0801aeb08ca9b7bf36c467d6476d0898f073d41b8512aa385c63271c64d36","abstract_canon_sha256":"e26bbb17f1d45ba5b99a42ba68eb2757597bbe74a8df728f6bd0e2c85d3e2dc6"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T07:00:09.711796Z","signature_b64":"wRvLnbpOVAU0oe7bbH7jQgbZM1ikYHGBzfi4+97u9m4SYBAcC/fssJ63EynFR5KBml4G1GlJQ6Chk1rYuOzyBg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"a594489aab72251afb223259425fcbf73f1397a0fa30b0fb625c18c086c3265f","last_reissued_at":"2026-07-05T07:00:09.711295Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T07:00:09.711295Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Tensor Programs VI: Feature Learning in Infinite-Depth Neural Networks","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cond-mat.dis-nn","math.PR"],"primary_cat":"cs.NE","authors_text":"Chen Zhu, Dingli Yu, Greg Yang, Soufiane Hayou","submitted_at":"2023-10-03T17:50:40Z","abstract_excerpt":"By classifying infinite-width neural networks and identifying the *optimal* limit, Tensor Programs IV and V demonstrated a universal way, called $\\mu$P, for *widthwise hyperparameter transfer*, i.e., predicting optimal hyperparameters of wide neural networks from narrow ones. Here we investigate the analogous classification for *depthwise parametrizations* of deep residual networks (resnets). We classify depthwise parametrizations of block multiplier and learning rate by their infinite-width-then-depth limits. In resnets where each block has only one layer, we identify a unique optimal paramet"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2310.02244","kind":"arxiv","version":5},"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/2310.02244/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":"2310.02244","created_at":"2026-07-05T07:00:09.711366+00:00"},{"alias_kind":"arxiv_version","alias_value":"2310.02244v5","created_at":"2026-07-05T07:00:09.711366+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2310.02244","created_at":"2026-07-05T07:00:09.711366+00:00"},{"alias_kind":"pith_short_12","alias_value":"UWKERGVLOISR","created_at":"2026-07-05T07:00:09.711366+00:00"},{"alias_kind":"pith_short_16","alias_value":"UWKERGVLOISRV6ZC","created_at":"2026-07-05T07:00:09.711366+00:00"},{"alias_kind":"pith_short_8","alias_value":"UWKERGVL","created_at":"2026-07-05T07:00:09.711366+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":6,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.25971","citing_title":"Improving Neural Network Training by Decoupling the Magnitude and Direction of Weight Vectors","ref_index":89,"is_internal_anchor":false},{"citing_arxiv_id":"2606.05610","citing_title":"Predictable Scaling Laws of Optimal Hyperparameters for LLM Continued Pre-training","ref_index":4,"is_internal_anchor":false},{"citing_arxiv_id":"2605.14200","citing_title":"How to Scale Mixture-of-Experts: From muP to the Maximally Scale-Stable Parameterization","ref_index":114,"is_internal_anchor":false},{"citing_arxiv_id":"2404.06395","citing_title":"MiniCPM: Unveiling the Potential of Small Language Models with Scalable Training Strategies","ref_index":47,"is_internal_anchor":false},{"citing_arxiv_id":"2604.21691","citing_title":"There Will Be a Scientific Theory of Deep Learning","ref_index":94,"is_internal_anchor":false},{"citing_arxiv_id":"2605.07815","citing_title":"OrScale: Orthogonalised Optimization with Layer-Wise Trust-Ratio Scaling","ref_index":21,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/UWKERGVLOISRV6ZCGJMUEX6L64","json":"https://pith.science/pith/UWKERGVLOISRV6ZCGJMUEX6L64.json","graph_json":"https://pith.science/api/pith-number/UWKERGVLOISRV6ZCGJMUEX6L64/graph.json","events_json":"https://pith.science/api/pith-number/UWKERGVLOISRV6ZCGJMUEX6L64/events.json","paper":"https://pith.science/paper/UWKERGVL"},"agent_actions":{"view_html":"https://pith.science/pith/UWKERGVLOISRV6ZCGJMUEX6L64","download_json":"https://pith.science/pith/UWKERGVLOISRV6ZCGJMUEX6L64.json","view_paper":"https://pith.science/paper/UWKERGVL","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2310.02244&json=true","fetch_graph":"https://pith.science/api/pith-number/UWKERGVLOISRV6ZCGJMUEX6L64/graph.json","fetch_events":"https://pith.science/api/pith-number/UWKERGVLOISRV6ZCGJMUEX6L64/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/UWKERGVLOISRV6ZCGJMUEX6L64/action/timestamp_anchor","attest_storage":"https://pith.science/pith/UWKERGVLOISRV6ZCGJMUEX6L64/action/storage_attestation","attest_author":"https://pith.science/pith/UWKERGVLOISRV6ZCGJMUEX6L64/action/author_attestation","sign_citation":"https://pith.science/pith/UWKERGVLOISRV6ZCGJMUEX6L64/action/citation_signature","submit_replication":"https://pith.science/pith/UWKERGVLOISRV6ZCGJMUEX6L64/action/replication_record"}},"created_at":"2026-07-05T07:00:09.711366+00:00","updated_at":"2026-07-05T07:00:09.711366+00:00"}