{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2020:AKH36ATNV6VVYB2AJ44SCYLMLJ","short_pith_number":"pith:AKH36ATN","schema_version":"1.0","canonical_sha256":"028fbf026dafab5c07404f3921616c5a62e9286d8821599fff984b7b8e2f223e","source":{"kind":"arxiv","id":"2004.10802","version":1},"attestation_state":"computed","paper":{"title":"A Neural Scaling Law from the Dimension of the Data Manifold","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["stat.ML"],"primary_cat":"cs.LG","authors_text":"Jared Kaplan, Utkarsh Sharma","submitted_at":"2020-04-22T19:16:06Z","abstract_excerpt":"When data is plentiful, the loss achieved by well-trained neural networks scales as a power-law $L \\propto N^{-\\alpha}$ in the number of network parameters $N$. This empirical scaling law holds for a wide variety of data modalities, and may persist over many orders of magnitude. The scaling law can be explained if neural models are effectively just performing regression on a data manifold of intrinsic dimension $d$. This simple theory predicts that the scaling exponents $\\alpha \\approx 4/d$ for cross-entropy and mean-squared error losses. We confirm the theory by independently measuring the in"},"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":"2004.10802","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2020-04-22T19:16:06Z","cross_cats_sorted":["stat.ML"],"title_canon_sha256":"cff65125676e57bf3c5f47cabaecfd8e7cf0961e60ab2cf5a9356ac803a6d750","abstract_canon_sha256":"76f701cb8eb63019b79081941f62ee4dd0856fdbcdf1bb6666f98ca6f817fc09"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T00:57:36.339492Z","signature_b64":"b1yF9VWYPCChx4UQ8WhcJbken/A1KccPwN1p87gciuZ6aRqDCsnArfla27tMYUlj8mlIWq8RtqFoGUIeeLStAg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"028fbf026dafab5c07404f3921616c5a62e9286d8821599fff984b7b8e2f223e","last_reissued_at":"2026-07-05T00:57:36.339046Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T00:57:36.339046Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"A Neural Scaling Law from the Dimension of the Data Manifold","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["stat.ML"],"primary_cat":"cs.LG","authors_text":"Jared Kaplan, Utkarsh Sharma","submitted_at":"2020-04-22T19:16:06Z","abstract_excerpt":"When data is plentiful, the loss achieved by well-trained neural networks scales as a power-law $L \\propto N^{-\\alpha}$ in the number of network parameters $N$. This empirical scaling law holds for a wide variety of data modalities, and may persist over many orders of magnitude. The scaling law can be explained if neural models are effectively just performing regression on a data manifold of intrinsic dimension $d$. This simple theory predicts that the scaling exponents $\\alpha \\approx 4/d$ for cross-entropy and mean-squared error losses. We confirm the theory by independently measuring the in"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2004.10802","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/2004.10802/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":"2004.10802","created_at":"2026-07-05T00:57:36.339102+00:00"},{"alias_kind":"arxiv_version","alias_value":"2004.10802v1","created_at":"2026-07-05T00:57:36.339102+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2004.10802","created_at":"2026-07-05T00:57:36.339102+00:00"},{"alias_kind":"pith_short_12","alias_value":"AKH36ATNV6VV","created_at":"2026-07-05T00:57:36.339102+00:00"},{"alias_kind":"pith_short_16","alias_value":"AKH36ATNV6VVYB2A","created_at":"2026-07-05T00:57:36.339102+00:00"},{"alias_kind":"pith_short_8","alias_value":"AKH36ATN","created_at":"2026-07-05T00:57:36.339102+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":14,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.26617","citing_title":"Sketched Linear Contrastive Learning: Approximation, Optimization, and Statistical Scaling","ref_index":8,"is_internal_anchor":false},{"citing_arxiv_id":"2606.19781","citing_title":"Towards Engineering Scaling Laws with Pretraining Data Composition","ref_index":28,"is_internal_anchor":false},{"citing_arxiv_id":"2606.01302","citing_title":"Structure and Scale in Simplicial Sequence Modelling","ref_index":60,"is_internal_anchor":false},{"citing_arxiv_id":"2605.24316","citing_title":"From One-Pass SGD to Data Reuse: Mini-Batch Scaling Laws in Sketched Linear Regression","ref_index":10,"is_internal_anchor":false},{"citing_arxiv_id":"2606.19781","citing_title":"Towards Engineering Scaling Laws with Pretraining Data Composition","ref_index":28,"is_internal_anchor":false},{"citing_arxiv_id":"2210.10760","citing_title":"Scaling Laws for Reward Model Overoptimization","ref_index":26,"is_internal_anchor":false},{"citing_arxiv_id":"2510.18900","citing_title":"Foundation Models for Discovery and Exploration in Chemical Space","ref_index":54,"is_internal_anchor":false},{"citing_arxiv_id":"2102.01293","citing_title":"Scaling Laws for Transfer","ref_index":191,"is_internal_anchor":false},{"citing_arxiv_id":"2405.14782","citing_title":"Lessons from the Trenches on Reproducible Evaluation of Language Models","ref_index":30,"is_internal_anchor":false},{"citing_arxiv_id":"2304.01373","citing_title":"Pythia: A Suite for Analyzing Large Language Models Across Training and Scaling","ref_index":123,"is_internal_anchor":false},{"citing_arxiv_id":"2010.14701","citing_title":"Scaling Laws for Autoregressive Generative Modeling","ref_index":22,"is_internal_anchor":false},{"citing_arxiv_id":"2404.19756","citing_title":"KAN: Kolmogorov-Arnold Networks","ref_index":24,"is_internal_anchor":false},{"citing_arxiv_id":"2112.00861","citing_title":"A General Language Assistant as a Laboratory for Alignment","ref_index":70,"is_internal_anchor":false},{"citing_arxiv_id":"2207.05221","citing_title":"Language Models (Mostly) Know What They Know","ref_index":128,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/AKH36ATNV6VVYB2AJ44SCYLMLJ","json":"https://pith.science/pith/AKH36ATNV6VVYB2AJ44SCYLMLJ.json","graph_json":"https://pith.science/api/pith-number/AKH36ATNV6VVYB2AJ44SCYLMLJ/graph.json","events_json":"https://pith.science/api/pith-number/AKH36ATNV6VVYB2AJ44SCYLMLJ/events.json","paper":"https://pith.science/paper/AKH36ATN"},"agent_actions":{"view_html":"https://pith.science/pith/AKH36ATNV6VVYB2AJ44SCYLMLJ","download_json":"https://pith.science/pith/AKH36ATNV6VVYB2AJ44SCYLMLJ.json","view_paper":"https://pith.science/paper/AKH36ATN","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2004.10802&json=true","fetch_graph":"https://pith.science/api/pith-number/AKH36ATNV6VVYB2AJ44SCYLMLJ/graph.json","fetch_events":"https://pith.science/api/pith-number/AKH36ATNV6VVYB2AJ44SCYLMLJ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/AKH36ATNV6VVYB2AJ44SCYLMLJ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/AKH36ATNV6VVYB2AJ44SCYLMLJ/action/storage_attestation","attest_author":"https://pith.science/pith/AKH36ATNV6VVYB2AJ44SCYLMLJ/action/author_attestation","sign_citation":"https://pith.science/pith/AKH36ATNV6VVYB2AJ44SCYLMLJ/action/citation_signature","submit_replication":"https://pith.science/pith/AKH36ATNV6VVYB2AJ44SCYLMLJ/action/replication_record"}},"created_at":"2026-07-05T00:57:36.339102+00:00","updated_at":"2026-07-05T00:57:36.339102+00:00"}