{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2022:VKN3CGCYPHXLOMFSYLJZLQUA4A","short_pith_number":"pith:VKN3CGCY","canonical_record":{"source":{"id":"2203.13240","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2022-03-24T17:50:46Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"00c34884b3f52bfadeda690ad441336d2f4ba296f52d2253aeaf9c540e83b0be","abstract_canon_sha256":"38dc068070a144c09b913ff95bbb5411f68f565918e819c6adbf32aa9694cad1"},"schema_version":"1.0"},"canonical_sha256":"aa9bb1185879eeb730b2c2d395c280e03c9d91af8c4f3ac1a00ae2865ceca172","source":{"kind":"arxiv","id":"2203.13240","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2203.13240","created_at":"2026-07-05T04:08:22Z"},{"alias_kind":"arxiv_version","alias_value":"2203.13240v1","created_at":"2026-07-05T04:08:22Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2203.13240","created_at":"2026-07-05T04:08:22Z"},{"alias_kind":"pith_short_12","alias_value":"VKN3CGCYPHXL","created_at":"2026-07-05T04:08:22Z"},{"alias_kind":"pith_short_16","alias_value":"VKN3CGCYPHXLOMFS","created_at":"2026-07-05T04:08:22Z"},{"alias_kind":"pith_short_8","alias_value":"VKN3CGCY","created_at":"2026-07-05T04:08:22Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2022:VKN3CGCYPHXLOMFSYLJZLQUA4A","target":"record","payload":{"canonical_record":{"source":{"id":"2203.13240","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2022-03-24T17:50:46Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"00c34884b3f52bfadeda690ad441336d2f4ba296f52d2253aeaf9c540e83b0be","abstract_canon_sha256":"38dc068070a144c09b913ff95bbb5411f68f565918e819c6adbf32aa9694cad1"},"schema_version":"1.0"},"canonical_sha256":"aa9bb1185879eeb730b2c2d395c280e03c9d91af8c4f3ac1a00ae2865ceca172","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T04:08:22.138927Z","signature_b64":"J+Tv1nPhaupk26wR5ksP8Uyqti/S1zpWVe+UHlGbw72nj9heLrHN+SpqlrkyuWuq3al5953yZNpe7mbqQF/RCw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"aa9bb1185879eeb730b2c2d395c280e03c9d91af8c4f3ac1a00ae2865ceca172","last_reissued_at":"2026-07-05T04:08:22.138380Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T04:08:22.138380Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2203.13240","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":"2026-07-05T04:08:22Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"dgYI2cuIYndCblU1C1TSDaVe7e/9SU2Y4JRJIyo/8GasfG+iAfGDCX9GSnpXoDxq4+WjyU3H+A6Zt2Ry3xG9Dg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-06T06:53:32.981083Z"},"content_sha256":"cb0c2c77c472af76270f9cad6ecbdf318d52dc0243d63717306b44126a54f3a4","schema_version":"1.0","event_id":"sha256:cb0c2c77c472af76270f9cad6ecbdf318d52dc0243d63717306b44126a54f3a4"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2022:VKN3CGCYPHXLOMFSYLJZLQUA4A","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Token Dropping for Efficient BERT Pretraining","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Denny Zhou, Le Hou, Richard Yuanzhe Pang, Tianyi Zhou, Xiaodan Song, Xinying Song, Yuexin Wu","submitted_at":"2022-03-24T17:50:46Z","abstract_excerpt":"Transformer-based models generally allocate the same amount of computation for each token in a given sequence. We develop a simple but effective \"token dropping\" method to accelerate the pretraining of transformer models, such as BERT, without degrading its performance on downstream tasks. In short, we drop unimportant tokens starting from an intermediate layer in the model to make the model focus on important tokens; the dropped tokens are later picked up by the last layer of the model so that the model still produces full-length sequences. We leverage the already built-in masked language mod"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2203.13240","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/2203.13240/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":"2026-07-05T04:08:22Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"y5JBWQyCYqNK+Ay2goE90W9Mvl3UNo2aNCGvfPqxkQF1t6JYbxMSFy242H8D+tcbhBIDjObY+MQT2zCjOScnDw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-06T06:53:32.982201Z"},"content_sha256":"754e4415c05e16ab07274e2f8ee2e0582d3b1e68b722375b360efb066697418b","schema_version":"1.0","event_id":"sha256:754e4415c05e16ab07274e2f8ee2e0582d3b1e68b722375b360efb066697418b"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/VKN3CGCYPHXLOMFSYLJZLQUA4A/bundle.json","state_url":"https://pith.science/pith/VKN3CGCYPHXLOMFSYLJZLQUA4A/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/VKN3CGCYPHXLOMFSYLJZLQUA4A/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":"2026-08-06T06:53:32Z","links":{"resolver":"https://pith.science/pith/VKN3CGCYPHXLOMFSYLJZLQUA4A","bundle":"https://pith.science/pith/VKN3CGCYPHXLOMFSYLJZLQUA4A/bundle.json","state":"https://pith.science/pith/VKN3CGCYPHXLOMFSYLJZLQUA4A/state.json","well_known_bundle":"https://pith.science/.well-known/pith/VKN3CGCYPHXLOMFSYLJZLQUA4A/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2022:VKN3CGCYPHXLOMFSYLJZLQUA4A","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":"38dc068070a144c09b913ff95bbb5411f68f565918e819c6adbf32aa9694cad1","cross_cats_sorted":["cs.LG"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2022-03-24T17:50:46Z","title_canon_sha256":"00c34884b3f52bfadeda690ad441336d2f4ba296f52d2253aeaf9c540e83b0be"},"schema_version":"1.0","source":{"id":"2203.13240","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2203.13240","created_at":"2026-07-05T04:08:22Z"},{"alias_kind":"arxiv_version","alias_value":"2203.13240v1","created_at":"2026-07-05T04:08:22Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2203.13240","created_at":"2026-07-05T04:08:22Z"},{"alias_kind":"pith_short_12","alias_value":"VKN3CGCYPHXL","created_at":"2026-07-05T04:08:22Z"},{"alias_kind":"pith_short_16","alias_value":"VKN3CGCYPHXLOMFS","created_at":"2026-07-05T04:08:22Z"},{"alias_kind":"pith_short_8","alias_value":"VKN3CGCY","created_at":"2026-07-05T04:08:22Z"}],"graph_snapshots":[{"event_id":"sha256:754e4415c05e16ab07274e2f8ee2e0582d3b1e68b722375b360efb066697418b","target":"graph","created_at":"2026-07-05T04:08:22Z","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/2203.13240/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Transformer-based models generally allocate the same amount of computation for each token in a given sequence. We develop a simple but effective \"token dropping\" method to accelerate the pretraining of transformer models, such as BERT, without degrading its performance on downstream tasks. In short, we drop unimportant tokens starting from an intermediate layer in the model to make the model focus on important tokens; the dropped tokens are later picked up by the last layer of the model so that the model still produces full-length sequences. We leverage the already built-in masked language mod","authors_text":"Denny Zhou, Le Hou, Richard Yuanzhe Pang, Tianyi Zhou, Xiaodan Song, Xinying Song, Yuexin Wu","cross_cats":["cs.LG"],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2022-03-24T17:50:46Z","title":"Token Dropping for Efficient BERT Pretraining"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2203.13240","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:cb0c2c77c472af76270f9cad6ecbdf318d52dc0243d63717306b44126a54f3a4","target":"record","created_at":"2026-07-05T04:08:22Z","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":"38dc068070a144c09b913ff95bbb5411f68f565918e819c6adbf32aa9694cad1","cross_cats_sorted":["cs.LG"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2022-03-24T17:50:46Z","title_canon_sha256":"00c34884b3f52bfadeda690ad441336d2f4ba296f52d2253aeaf9c540e83b0be"},"schema_version":"1.0","source":{"id":"2203.13240","kind":"arxiv","version":1}},"canonical_sha256":"aa9bb1185879eeb730b2c2d395c280e03c9d91af8c4f3ac1a00ae2865ceca172","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"aa9bb1185879eeb730b2c2d395c280e03c9d91af8c4f3ac1a00ae2865ceca172","first_computed_at":"2026-07-05T04:08:22.138380Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T04:08:22.138380Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"J+Tv1nPhaupk26wR5ksP8Uyqti/S1zpWVe+UHlGbw72nj9heLrHN+SpqlrkyuWuq3al5953yZNpe7mbqQF/RCw==","signature_status":"signed_v1","signed_at":"2026-07-05T04:08:22.138927Z","signed_message":"canonical_sha256_bytes"},"source_id":"2203.13240","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:cb0c2c77c472af76270f9cad6ecbdf318d52dc0243d63717306b44126a54f3a4","sha256:754e4415c05e16ab07274e2f8ee2e0582d3b1e68b722375b360efb066697418b"],"state_sha256":"c494c6393037b375fecfe740b73b343dc66a502e6997b75891ca72c18a95ad32"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"JU8YzqykNGor0q3hZ4bR1RSlcUHESDAA/IYFre6RufNByinzGAsvsGU2ohic5j3jsu4t/dD69kMOwxD9QT6bAQ==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-06T06:53:32.988679Z","bundle_sha256":"bf1fa8211e2c30f8bf6d6b27ca968bdc6560b111c27f1ed2a3b0fcaa30627574"}}