{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2022:6OSDXP2TIH6X5PMBFTUZPZKYZ2","short_pith_number":"pith:6OSDXP2T","schema_version":"1.0","canonical_sha256":"f3a43bbf5341fd7ebd812ce997e558ce9a65f2bc106839832d876d37c5e27ad2","source":{"kind":"arxiv","id":"2203.06211","version":1},"attestation_state":"computed","paper":{"title":"Staged Training for Transformer Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Iz Beltagy, Jesse Dodge, Kurt Keutzer, Matthew Peters, Pete Walsh, Sheng Shen","submitted_at":"2022-03-11T19:05:42Z","abstract_excerpt":"The current standard approach to scaling transformer language models trains each model size from a different random initialization. As an alternative, we consider a staged training setup that begins with a small model and incrementally increases the amount of compute used for training by applying a \"growth operator\" to increase the model depth and width. By initializing each stage with the output of the previous one, the training process effectively re-uses the compute from prior stages and becomes more efficient. Our growth operators each take as input the entire training state (including mod"},"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":"2203.06211","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2022-03-11T19:05:42Z","cross_cats_sorted":[],"title_canon_sha256":"7ca3fb05b8ff4a694f4e91f018768a62ca26a23dbc6ff53396a5e5f1a7fea4bd","abstract_canon_sha256":"2d1a4338783bf8913c8e5dbd6fc12f3622f3e3e47c3ba3d56ab109d2185cc2eb"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T04:04:48.268619Z","signature_b64":"9q9tU0Jrfcd54/EDjsOiTrG5q1hYqhxbQxcOvYPBwdKr8ZJTi+4+j0e1VliYpieGo9gueQoSWXQ/AThznwzWCQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"f3a43bbf5341fd7ebd812ce997e558ce9a65f2bc106839832d876d37c5e27ad2","last_reissued_at":"2026-07-05T04:04:48.268146Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T04:04:48.268146Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Staged Training for Transformer Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Iz Beltagy, Jesse Dodge, Kurt Keutzer, Matthew Peters, Pete Walsh, Sheng Shen","submitted_at":"2022-03-11T19:05:42Z","abstract_excerpt":"The current standard approach to scaling transformer language models trains each model size from a different random initialization. As an alternative, we consider a staged training setup that begins with a small model and incrementally increases the amount of compute used for training by applying a \"growth operator\" to increase the model depth and width. By initializing each stage with the output of the previous one, the training process effectively re-uses the compute from prior stages and becomes more efficient. Our growth operators each take as input the entire training state (including mod"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2203.06211","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/2203.06211/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":"2203.06211","created_at":"2026-07-05T04:04:48.268209+00:00"},{"alias_kind":"arxiv_version","alias_value":"2203.06211v1","created_at":"2026-07-05T04:04:48.268209+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2203.06211","created_at":"2026-07-05T04:04:48.268209+00:00"},{"alias_kind":"pith_short_12","alias_value":"6OSDXP2TIH6X","created_at":"2026-07-05T04:04:48.268209+00:00"},{"alias_kind":"pith_short_16","alias_value":"6OSDXP2TIH6X5PMB","created_at":"2026-07-05T04:04:48.268209+00:00"},{"alias_kind":"pith_short_8","alias_value":"6OSDXP2T","created_at":"2026-07-05T04:04:48.268209+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/6OSDXP2TIH6X5PMBFTUZPZKYZ2","json":"https://pith.science/pith/6OSDXP2TIH6X5PMBFTUZPZKYZ2.json","graph_json":"https://pith.science/api/pith-number/6OSDXP2TIH6X5PMBFTUZPZKYZ2/graph.json","events_json":"https://pith.science/api/pith-number/6OSDXP2TIH6X5PMBFTUZPZKYZ2/events.json","paper":"https://pith.science/paper/6OSDXP2T"},"agent_actions":{"view_html":"https://pith.science/pith/6OSDXP2TIH6X5PMBFTUZPZKYZ2","download_json":"https://pith.science/pith/6OSDXP2TIH6X5PMBFTUZPZKYZ2.json","view_paper":"https://pith.science/paper/6OSDXP2T","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2203.06211&json=true","fetch_graph":"https://pith.science/api/pith-number/6OSDXP2TIH6X5PMBFTUZPZKYZ2/graph.json","fetch_events":"https://pith.science/api/pith-number/6OSDXP2TIH6X5PMBFTUZPZKYZ2/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/6OSDXP2TIH6X5PMBFTUZPZKYZ2/action/timestamp_anchor","attest_storage":"https://pith.science/pith/6OSDXP2TIH6X5PMBFTUZPZKYZ2/action/storage_attestation","attest_author":"https://pith.science/pith/6OSDXP2TIH6X5PMBFTUZPZKYZ2/action/author_attestation","sign_citation":"https://pith.science/pith/6OSDXP2TIH6X5PMBFTUZPZKYZ2/action/citation_signature","submit_replication":"https://pith.science/pith/6OSDXP2TIH6X5PMBFTUZPZKYZ2/action/replication_record"}},"created_at":"2026-07-05T04:04:48.268209+00:00","updated_at":"2026-07-05T04:04:48.268209+00:00"}