{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2022:6OSDXP2TIH6X5PMBFTUZPZKYZ2","short_pith_number":"pith:6OSDXP2T","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"},"canonical_sha256":"f3a43bbf5341fd7ebd812ce997e558ce9a65f2bc106839832d876d37c5e27ad2","source":{"kind":"arxiv","id":"2203.06211","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2203.06211","created_at":"2026-07-05T04:04:48Z"},{"alias_kind":"arxiv_version","alias_value":"2203.06211v1","created_at":"2026-07-05T04:04:48Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2203.06211","created_at":"2026-07-05T04:04:48Z"},{"alias_kind":"pith_short_12","alias_value":"6OSDXP2TIH6X","created_at":"2026-07-05T04:04:48Z"},{"alias_kind":"pith_short_16","alias_value":"6OSDXP2TIH6X5PMB","created_at":"2026-07-05T04:04:48Z"},{"alias_kind":"pith_short_8","alias_value":"6OSDXP2T","created_at":"2026-07-05T04:04:48Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2022:6OSDXP2TIH6X5PMBFTUZPZKYZ2","target":"record","payload":{"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"},"canonical_sha256":"f3a43bbf5341fd7ebd812ce997e558ce9a65f2bc106839832d876d37c5e27ad2","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"},"source_kind":"arxiv","source_id":"2203.06211","source_version":1,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T04:04:48Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"6w9PA1xNSpn60MJCz+9V31DFnh8SB2kzs3fYuoIrrCG+AwPAjocHp/bDF8EjRB0yyXUhkM47wBxFcxo5fniUAA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-07T17:59:26.469357Z"},"content_sha256":"4d904ecd9f33c31615a0698cba64ab135b5d1016c2ff0280c9c4fbf820f13a54","schema_version":"1.0","event_id":"sha256:4d904ecd9f33c31615a0698cba64ab135b5d1016c2ff0280c9c4fbf820f13a54"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2022:6OSDXP2TIH6X5PMBFTUZPZKYZ2","target":"graph","payload":{"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"},"verdict_id":null},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T04:04:48Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"0VOs3MGTKvai23OkF/EAo4TM2+csu4+Cn8pYl44xDYKDNX9IyQqiqkJOkSuYLNRTuo7snXr/trg6OKf4Zi+RDw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-07T17:59:26.469867Z"},"content_sha256":"4225ad289c9377979b95435630a5bdb44380c829ce46771138ea71e6ca8a8ec2","schema_version":"1.0","event_id":"sha256:4225ad289c9377979b95435630a5bdb44380c829ce46771138ea71e6ca8a8ec2"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/6OSDXP2TIH6X5PMBFTUZPZKYZ2/bundle.json","state_url":"https://pith.science/pith/6OSDXP2TIH6X5PMBFTUZPZKYZ2/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/6OSDXP2TIH6X5PMBFTUZPZKYZ2/bundle.json","status":"primary"}],"public_keys":[{"key_id":"pith-v1-2026-05","algorithm":"ed25519","format":"raw","public_key_b64":"stVStoiQhXFxp4s2pdzPNoqVNBMojDU/fJ2db5S3CbM=","public_key_hex":"b2d552b68890857171a78b36a5dccf368a953413288c353f7c9d9d6f94b709b3","fingerprint_sha256_b32_first128bits":"RVFV5Z2OI2J3ZUO7ERDEBCYNKS","fingerprint_sha256_hex":"8d4b5ee74e4693bcd1df2446408b0d54","rotates_at":null,"url":"https://pith.science/pith-signing-key.json","notes":"Pith uses this Ed25519 key to sign canonical record SHA-256 digests. Verify with: ed25519_verify(public_key, message=canonical_sha256_bytes, signature=base64decode(signature_b64))."}],"merge_version":"pith-open-graph-merge-v1","built_at":"2026-08-07T17:59:26Z","links":{"resolver":"https://pith.science/pith/6OSDXP2TIH6X5PMBFTUZPZKYZ2","bundle":"https://pith.science/pith/6OSDXP2TIH6X5PMBFTUZPZKYZ2/bundle.json","state":"https://pith.science/pith/6OSDXP2TIH6X5PMBFTUZPZKYZ2/state.json","well_known_bundle":"https://pith.science/.well-known/pith/6OSDXP2TIH6X5PMBFTUZPZKYZ2/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2022:6OSDXP2TIH6X5PMBFTUZPZKYZ2","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"2d1a4338783bf8913c8e5dbd6fc12f3622f3e3e47c3ba3d56ab109d2185cc2eb","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2022-03-11T19:05:42Z","title_canon_sha256":"7ca3fb05b8ff4a694f4e91f018768a62ca26a23dbc6ff53396a5e5f1a7fea4bd"},"schema_version":"1.0","source":{"id":"2203.06211","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2203.06211","created_at":"2026-07-05T04:04:48Z"},{"alias_kind":"arxiv_version","alias_value":"2203.06211v1","created_at":"2026-07-05T04:04:48Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2203.06211","created_at":"2026-07-05T04:04:48Z"},{"alias_kind":"pith_short_12","alias_value":"6OSDXP2TIH6X","created_at":"2026-07-05T04:04:48Z"},{"alias_kind":"pith_short_16","alias_value":"6OSDXP2TIH6X5PMB","created_at":"2026-07-05T04:04:48Z"},{"alias_kind":"pith_short_8","alias_value":"6OSDXP2T","created_at":"2026-07-05T04:04:48Z"}],"graph_snapshots":[{"event_id":"sha256:4225ad289c9377979b95435630a5bdb44380c829ce46771138ea71e6ca8a8ec2","target":"graph","created_at":"2026-07-05T04:04:48Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/2203.06211/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"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","authors_text":"Iz Beltagy, Jesse Dodge, Kurt Keutzer, Matthew Peters, Pete Walsh, Sheng Shen","cross_cats":[],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2022-03-11T19:05:42Z","title":"Staged Training for Transformer Language Models"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2203.06211","kind":"arxiv","version":1},"verdict":{"created_at":null,"id":null,"model_set":{},"one_line_summary":"","pipeline_version":null,"pith_extraction_headline":"","strongest_claim":"","weakest_assumption":""}},"verdict_id":null}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:4d904ecd9f33c31615a0698cba64ab135b5d1016c2ff0280c9c4fbf820f13a54","target":"record","created_at":"2026-07-05T04:04:48Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"2d1a4338783bf8913c8e5dbd6fc12f3622f3e3e47c3ba3d56ab109d2185cc2eb","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2022-03-11T19:05:42Z","title_canon_sha256":"7ca3fb05b8ff4a694f4e91f018768a62ca26a23dbc6ff53396a5e5f1a7fea4bd"},"schema_version":"1.0","source":{"id":"2203.06211","kind":"arxiv","version":1}},"canonical_sha256":"f3a43bbf5341fd7ebd812ce997e558ce9a65f2bc106839832d876d37c5e27ad2","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"f3a43bbf5341fd7ebd812ce997e558ce9a65f2bc106839832d876d37c5e27ad2","first_computed_at":"2026-07-05T04:04:48.268146Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T04:04:48.268146Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"9q9tU0Jrfcd54/EDjsOiTrG5q1hYqhxbQxcOvYPBwdKr8ZJTi+4+j0e1VliYpieGo9gueQoSWXQ/AThznwzWCQ==","signature_status":"signed_v1","signed_at":"2026-07-05T04:04:48.268619Z","signed_message":"canonical_sha256_bytes"},"source_id":"2203.06211","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:4d904ecd9f33c31615a0698cba64ab135b5d1016c2ff0280c9c4fbf820f13a54","sha256:4225ad289c9377979b95435630a5bdb44380c829ce46771138ea71e6ca8a8ec2"],"state_sha256":"5bcd675f6d2e43037ea311953e6ce2c1d798eb57141bab2c7a68c8d77499e876"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"0Ban5fvHYW2AAzhMLq91++KAhZZs9jqj7RPmYOx3MZ/WnPMbpre9Oqo+5OyMmo6DjS+N7liGXfGzvsVCJ731BA==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-07T17:59:26.474647Z","bundle_sha256":"f1346cf7e4ed1c184be56e415c7a50be9c03453dfa18fdebaf9d3e45b438dbe5"}}