{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2022:SHRPS3F6QA5SGNHZ3UYAZXLCCL","short_pith_number":"pith:SHRPS3F6","schema_version":"1.0","canonical_sha256":"91e2f96cbe803b2334f9dd300cdd6212e5b179aff937ea977f196737cd694f4b","source":{"kind":"arxiv","id":"2202.05924","version":2},"attestation_state":"computed","paper":{"title":"Compute Trends Across Three Eras of Machine Learning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CY"],"primary_cat":"cs.LG","authors_text":"Anson Ho, Jaime Sevilla, Lennart Heim, Marius Hobbhahn, Pablo Villalobos, Tamay Besiroglu","submitted_at":"2022-02-11T22:42:47Z","abstract_excerpt":"Compute, data, and algorithmic advances are the three fundamental factors that guide the progress of modern Machine Learning (ML). In this paper we study trends in the most readily quantified factor - compute. We show that before 2010 training compute grew in line with Moore's law, doubling roughly every 20 months. Since the advent of Deep Learning in the early 2010s, the scaling of training compute has accelerated, doubling approximately every 6 months. In late 2015, a new trend emerged as firms developed large-scale ML models with 10 to 100-fold larger requirements in training compute. Based"},"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":"2202.05924","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2022-02-11T22:42:47Z","cross_cats_sorted":["cs.AI","cs.CY"],"title_canon_sha256":"988230475cf54bba2711a8aa46481d8197e03ce55378a9254ed81c0100990d5d","abstract_canon_sha256":"88cc1617ffdb3edb31cfc850abbd8b0a2b0ec6d1460f2c8213660db0ce45dee0"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T07:14:05.006420Z","signature_b64":"JHBw6e1it+TcpG8wwXGR/98gUokvpiJma1wIhNrxhuTtt56lF/Gsjiwb+Ny0EUXeSdPLBr9VfmRoe2wZhiEGAA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"91e2f96cbe803b2334f9dd300cdd6212e5b179aff937ea977f196737cd694f4b","last_reissued_at":"2026-07-05T07:14:05.005919Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T07:14:05.005919Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Compute Trends Across Three Eras of Machine Learning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CY"],"primary_cat":"cs.LG","authors_text":"Anson Ho, Jaime Sevilla, Lennart Heim, Marius Hobbhahn, Pablo Villalobos, Tamay Besiroglu","submitted_at":"2022-02-11T22:42:47Z","abstract_excerpt":"Compute, data, and algorithmic advances are the three fundamental factors that guide the progress of modern Machine Learning (ML). In this paper we study trends in the most readily quantified factor - compute. We show that before 2010 training compute grew in line with Moore's law, doubling roughly every 20 months. Since the advent of Deep Learning in the early 2010s, the scaling of training compute has accelerated, doubling approximately every 6 months. In late 2015, a new trend emerged as firms developed large-scale ML models with 10 to 100-fold larger requirements in training compute. Based"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2202.05924","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/2202.05924/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":"2202.05924","created_at":"2026-07-05T07:14:05.006024+00:00"},{"alias_kind":"arxiv_version","alias_value":"2202.05924v2","created_at":"2026-07-05T07:14:05.006024+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2202.05924","created_at":"2026-07-05T07:14:05.006024+00:00"},{"alias_kind":"pith_short_12","alias_value":"SHRPS3F6QA5S","created_at":"2026-07-05T07:14:05.006024+00:00"},{"alias_kind":"pith_short_16","alias_value":"SHRPS3F6QA5SGNHZ","created_at":"2026-07-05T07:14:05.006024+00:00"},{"alias_kind":"pith_short_8","alias_value":"SHRPS3F6","created_at":"2026-07-05T07:14:05.006024+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":4,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.07553","citing_title":"MedicalRec: Medical recommender system for image classification without retraining","ref_index":3,"is_internal_anchor":false},{"citing_arxiv_id":"2512.03915","citing_title":"A Theoretical Framework for Auxiliary-Loss-Free Load Balancing of Sparse Mixture-of-Experts in Large-Scale AI Models","ref_index":12,"is_internal_anchor":false},{"citing_arxiv_id":"2601.14053","citing_title":"LLMOrbit: A Circular Taxonomy of Large Language Models -From Scaling Walls to Agentic AI Systems","ref_index":136,"is_internal_anchor":false},{"citing_arxiv_id":"2605.11733","citing_title":"Position: LLM Inference Should Be Evaluated as Energy-to-Token Production","ref_index":37,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/SHRPS3F6QA5SGNHZ3UYAZXLCCL","json":"https://pith.science/pith/SHRPS3F6QA5SGNHZ3UYAZXLCCL.json","graph_json":"https://pith.science/api/pith-number/SHRPS3F6QA5SGNHZ3UYAZXLCCL/graph.json","events_json":"https://pith.science/api/pith-number/SHRPS3F6QA5SGNHZ3UYAZXLCCL/events.json","paper":"https://pith.science/paper/SHRPS3F6"},"agent_actions":{"view_html":"https://pith.science/pith/SHRPS3F6QA5SGNHZ3UYAZXLCCL","download_json":"https://pith.science/pith/SHRPS3F6QA5SGNHZ3UYAZXLCCL.json","view_paper":"https://pith.science/paper/SHRPS3F6","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2202.05924&json=true","fetch_graph":"https://pith.science/api/pith-number/SHRPS3F6QA5SGNHZ3UYAZXLCCL/graph.json","fetch_events":"https://pith.science/api/pith-number/SHRPS3F6QA5SGNHZ3UYAZXLCCL/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/SHRPS3F6QA5SGNHZ3UYAZXLCCL/action/timestamp_anchor","attest_storage":"https://pith.science/pith/SHRPS3F6QA5SGNHZ3UYAZXLCCL/action/storage_attestation","attest_author":"https://pith.science/pith/SHRPS3F6QA5SGNHZ3UYAZXLCCL/action/author_attestation","sign_citation":"https://pith.science/pith/SHRPS3F6QA5SGNHZ3UYAZXLCCL/action/citation_signature","submit_replication":"https://pith.science/pith/SHRPS3F6QA5SGNHZ3UYAZXLCCL/action/replication_record"}},"created_at":"2026-07-05T07:14:05.006024+00:00","updated_at":"2026-07-05T07:14:05.006024+00:00"}