{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2020:FGGZ66MQDNBVODUWJFWZK5EPTI","short_pith_number":"pith:FGGZ66MQ","canonical_record":{"source":{"id":"2004.03844","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2020-04-08T07:09:59Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"4810b4d4b0004375988ab9a937510ac11c77578362e0cc2d348c1ba82587a584","abstract_canon_sha256":"eafb742500653e53d08c05c8d5f1b22859def4aa7defbf9d44a4f4809eda1e28"},"schema_version":"1.0"},"canonical_sha256":"298d9f79901b43570e96496d95748f9a075f677365b199de8685164bf9ef40c1","source":{"kind":"arxiv","id":"2004.03844","version":3},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2004.03844","created_at":"2026-07-05T04:48:05Z"},{"alias_kind":"arxiv_version","alias_value":"2004.03844v3","created_at":"2026-07-05T04:48:05Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2004.03844","created_at":"2026-07-05T04:48:05Z"},{"alias_kind":"pith_short_12","alias_value":"FGGZ66MQDNBV","created_at":"2026-07-05T04:48:05Z"},{"alias_kind":"pith_short_16","alias_value":"FGGZ66MQDNBVODUW","created_at":"2026-07-05T04:48:05Z"},{"alias_kind":"pith_short_8","alias_value":"FGGZ66MQ","created_at":"2026-07-05T04:48:05Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2020:FGGZ66MQDNBVODUWJFWZK5EPTI","target":"record","payload":{"canonical_record":{"source":{"id":"2004.03844","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2020-04-08T07:09:59Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"4810b4d4b0004375988ab9a937510ac11c77578362e0cc2d348c1ba82587a584","abstract_canon_sha256":"eafb742500653e53d08c05c8d5f1b22859def4aa7defbf9d44a4f4809eda1e28"},"schema_version":"1.0"},"canonical_sha256":"298d9f79901b43570e96496d95748f9a075f677365b199de8685164bf9ef40c1","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T04:48:05.663689Z","signature_b64":"6IXcG13UH36va5mu08D/D0PfJEIbHrxhmTcpFCWuy7WZ43UQVIbycUhAFS16HDSUCIP02jUhaYba6NmO5kqwBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"298d9f79901b43570e96496d95748f9a075f677365b199de8685164bf9ef40c1","last_reissued_at":"2026-07-05T04:48:05.663289Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T04:48:05.663289Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2004.03844","source_version":3,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T04:48:05Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"xj0OQUEVY6MeVMKD7t+/qRMw3Rab9/cBG2IIGC6RrqOr6W4nCR1CsiVQ1+DT4HuqrFTGG2iA5d35YNnBi6tGCQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-04T09:42:03.561109Z"},"content_sha256":"c56278ae5caa7538f2d197cb26a3b687a7db6418d0abc634d74dfa3569439618","schema_version":"1.0","event_id":"sha256:c56278ae5caa7538f2d197cb26a3b687a7db6418d0abc634d74dfa3569439618"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2020:FGGZ66MQDNBVODUWJFWZK5EPTI","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"On the Effect of Dropping Layers of Pre-trained Transformer Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Fahim Dalvi, Hassan Sajjad, Nadir Durrani, Preslav Nakov","submitted_at":"2020-04-08T07:09:59Z","abstract_excerpt":"Transformer-based NLP models are trained using hundreds of millions or even billions of parameters, limiting their applicability in computationally constrained environments. While the number of parameters generally correlates with performance, it is not clear whether the entire network is required for a downstream task. Motivated by the recent work on pruning and distilling pre-trained models, we explore strategies to drop layers in pre-trained models, and observe the effect of pruning on downstream GLUE tasks. We were able to prune BERT, RoBERTa and XLNet models up to 40%, while maintaining u"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2004.03844","kind":"arxiv","version":3},"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/2004.03844/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":"2026-07-05T04:48:05Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"bViSBTf9BwmDeqnoRQWwiZLjyyUsOBpYL79sw0D+wTnxXflY9KKBP8fCdD8aPtNcrjOM3JgCHHTq+ZdMFNHkBQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-04T09:42:03.561643Z"},"content_sha256":"f9844ac0f0eb88f2708396f268f53ee0a42c4727b68b26fb1b0b873b1bc2db3b","schema_version":"1.0","event_id":"sha256:f9844ac0f0eb88f2708396f268f53ee0a42c4727b68b26fb1b0b873b1bc2db3b"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/FGGZ66MQDNBVODUWJFWZK5EPTI/bundle.json","state_url":"https://pith.science/pith/FGGZ66MQDNBVODUWJFWZK5EPTI/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/FGGZ66MQDNBVODUWJFWZK5EPTI/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":"2026-08-04T09:42:03Z","links":{"resolver":"https://pith.science/pith/FGGZ66MQDNBVODUWJFWZK5EPTI","bundle":"https://pith.science/pith/FGGZ66MQDNBVODUWJFWZK5EPTI/bundle.json","state":"https://pith.science/pith/FGGZ66MQDNBVODUWJFWZK5EPTI/state.json","well_known_bundle":"https://pith.science/.well-known/pith/FGGZ66MQDNBVODUWJFWZK5EPTI/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2020:FGGZ66MQDNBVODUWJFWZK5EPTI","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":"eafb742500653e53d08c05c8d5f1b22859def4aa7defbf9d44a4f4809eda1e28","cross_cats_sorted":["cs.LG"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2020-04-08T07:09:59Z","title_canon_sha256":"4810b4d4b0004375988ab9a937510ac11c77578362e0cc2d348c1ba82587a584"},"schema_version":"1.0","source":{"id":"2004.03844","kind":"arxiv","version":3}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2004.03844","created_at":"2026-07-05T04:48:05Z"},{"alias_kind":"arxiv_version","alias_value":"2004.03844v3","created_at":"2026-07-05T04:48:05Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2004.03844","created_at":"2026-07-05T04:48:05Z"},{"alias_kind":"pith_short_12","alias_value":"FGGZ66MQDNBV","created_at":"2026-07-05T04:48:05Z"},{"alias_kind":"pith_short_16","alias_value":"FGGZ66MQDNBVODUW","created_at":"2026-07-05T04:48:05Z"},{"alias_kind":"pith_short_8","alias_value":"FGGZ66MQ","created_at":"2026-07-05T04:48:05Z"}],"graph_snapshots":[{"event_id":"sha256:f9844ac0f0eb88f2708396f268f53ee0a42c4727b68b26fb1b0b873b1bc2db3b","target":"graph","created_at":"2026-07-05T04:48:05Z","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/2004.03844/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Transformer-based NLP models are trained using hundreds of millions or even billions of parameters, limiting their applicability in computationally constrained environments. While the number of parameters generally correlates with performance, it is not clear whether the entire network is required for a downstream task. Motivated by the recent work on pruning and distilling pre-trained models, we explore strategies to drop layers in pre-trained models, and observe the effect of pruning on downstream GLUE tasks. We were able to prune BERT, RoBERTa and XLNet models up to 40%, while maintaining u","authors_text":"Fahim Dalvi, Hassan Sajjad, Nadir Durrani, Preslav Nakov","cross_cats":["cs.LG"],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2020-04-08T07:09:59Z","title":"On the Effect of Dropping Layers of Pre-trained Transformer Models"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2004.03844","kind":"arxiv","version":3},"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:c56278ae5caa7538f2d197cb26a3b687a7db6418d0abc634d74dfa3569439618","target":"record","created_at":"2026-07-05T04:48:05Z","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":"eafb742500653e53d08c05c8d5f1b22859def4aa7defbf9d44a4f4809eda1e28","cross_cats_sorted":["cs.LG"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2020-04-08T07:09:59Z","title_canon_sha256":"4810b4d4b0004375988ab9a937510ac11c77578362e0cc2d348c1ba82587a584"},"schema_version":"1.0","source":{"id":"2004.03844","kind":"arxiv","version":3}},"canonical_sha256":"298d9f79901b43570e96496d95748f9a075f677365b199de8685164bf9ef40c1","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"298d9f79901b43570e96496d95748f9a075f677365b199de8685164bf9ef40c1","first_computed_at":"2026-07-05T04:48:05.663289Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T04:48:05.663289Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"6IXcG13UH36va5mu08D/D0PfJEIbHrxhmTcpFCWuy7WZ43UQVIbycUhAFS16HDSUCIP02jUhaYba6NmO5kqwBA==","signature_status":"signed_v1","signed_at":"2026-07-05T04:48:05.663689Z","signed_message":"canonical_sha256_bytes"},"source_id":"2004.03844","source_kind":"arxiv","source_version":3}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:c56278ae5caa7538f2d197cb26a3b687a7db6418d0abc634d74dfa3569439618","sha256:f9844ac0f0eb88f2708396f268f53ee0a42c4727b68b26fb1b0b873b1bc2db3b"],"state_sha256":"af434b848575dc002cf2ae9804fb77d67266fdc981aa6fb39c9c72dece53e091"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"6MI+Zbnjwq/ouItei3aNG9zUPzjuDR6zOufVM0rOaq0i7YaJcaJQYdFdsh0ifr4aXYLUSs7OHvmVI3WgLDwYAw==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-04T09:42:03.567883Z","bundle_sha256":"c282cc53845efe6f373205f0e409e716280f0bf27c50f277111607001c40092a"}}