{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2022:BR5ILJPE7DL32WB55O6FOC32BV","short_pith_number":"pith:BR5ILJPE","schema_version":"1.0","canonical_sha256":"0c7a85a5e4f8d7bd583debbc570b7a0d6fa365ae5f794b187dd733b4da7faf1b","source":{"kind":"arxiv","id":"2211.05610","version":2},"attestation_state":"computed","paper":{"title":"BERT on a Data Diet: Finding Important Examples by Gradient-Based Pruning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Ali Modarressi, Ehsan Aghazadeh, Mohammad Taher Pilehvar, Mohsen Fayyaz, Samira Ebrahimi Kahou, Yadollah Yaghoobzadeh","submitted_at":"2022-11-10T14:37:23Z","abstract_excerpt":"Current pre-trained language models rely on large datasets for achieving state-of-the-art performance. However, past research has shown that not all examples in a dataset are equally important during training. In fact, it is sometimes possible to prune a considerable fraction of the training set while maintaining the test performance. Established on standard vision benchmarks, two gradient-based scoring metrics for finding important examples are GraNd and its estimated version, EL2N. In this work, we employ these two metrics for the first time in NLP. We demonstrate that these metrics need to "},"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":"2211.05610","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2022-11-10T14:37:23Z","cross_cats_sorted":[],"title_canon_sha256":"c27204585193692ecbdd2627e080a3f0dbf563761ec74c45bf777f686942aaea","abstract_canon_sha256":"4786f6a7a4d60caa0102701aac87cad6bf2749c05fdcf2488fedd35e979653f8"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T05:19:56.003449Z","signature_b64":"aT8oNq3SXFYlB1SYgab2vOdk1+VvS6DBxPo4nI69uT7WUrQqgheNk9FsXLdbDxidZNy/BFIQs0wvO4Leae6mBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"0c7a85a5e4f8d7bd583debbc570b7a0d6fa365ae5f794b187dd733b4da7faf1b","last_reissued_at":"2026-07-05T05:19:56.003017Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T05:19:56.003017Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"BERT on a Data Diet: Finding Important Examples by Gradient-Based Pruning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Ali Modarressi, Ehsan Aghazadeh, Mohammad Taher Pilehvar, Mohsen Fayyaz, Samira Ebrahimi Kahou, Yadollah Yaghoobzadeh","submitted_at":"2022-11-10T14:37:23Z","abstract_excerpt":"Current pre-trained language models rely on large datasets for achieving state-of-the-art performance. However, past research has shown that not all examples in a dataset are equally important during training. In fact, it is sometimes possible to prune a considerable fraction of the training set while maintaining the test performance. Established on standard vision benchmarks, two gradient-based scoring metrics for finding important examples are GraNd and its estimated version, EL2N. In this work, we employ these two metrics for the first time in NLP. We demonstrate that these metrics need to "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2211.05610","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/2211.05610/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":"2211.05610","created_at":"2026-07-05T05:19:56.003073+00:00"},{"alias_kind":"arxiv_version","alias_value":"2211.05610v2","created_at":"2026-07-05T05:19:56.003073+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2211.05610","created_at":"2026-07-05T05:19:56.003073+00:00"},{"alias_kind":"pith_short_12","alias_value":"BR5ILJPE7DL3","created_at":"2026-07-05T05:19:56.003073+00:00"},{"alias_kind":"pith_short_16","alias_value":"BR5ILJPE7DL32WB5","created_at":"2026-07-05T05:19:56.003073+00:00"},{"alias_kind":"pith_short_8","alias_value":"BR5ILJPE","created_at":"2026-07-05T05:19:56.003073+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2505.18458","citing_title":"A Survey of LLM $\\times$ DATA","ref_index":137,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/BR5ILJPE7DL32WB55O6FOC32BV","json":"https://pith.science/pith/BR5ILJPE7DL32WB55O6FOC32BV.json","graph_json":"https://pith.science/api/pith-number/BR5ILJPE7DL32WB55O6FOC32BV/graph.json","events_json":"https://pith.science/api/pith-number/BR5ILJPE7DL32WB55O6FOC32BV/events.json","paper":"https://pith.science/paper/BR5ILJPE"},"agent_actions":{"view_html":"https://pith.science/pith/BR5ILJPE7DL32WB55O6FOC32BV","download_json":"https://pith.science/pith/BR5ILJPE7DL32WB55O6FOC32BV.json","view_paper":"https://pith.science/paper/BR5ILJPE","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2211.05610&json=true","fetch_graph":"https://pith.science/api/pith-number/BR5ILJPE7DL32WB55O6FOC32BV/graph.json","fetch_events":"https://pith.science/api/pith-number/BR5ILJPE7DL32WB55O6FOC32BV/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/BR5ILJPE7DL32WB55O6FOC32BV/action/timestamp_anchor","attest_storage":"https://pith.science/pith/BR5ILJPE7DL32WB55O6FOC32BV/action/storage_attestation","attest_author":"https://pith.science/pith/BR5ILJPE7DL32WB55O6FOC32BV/action/author_attestation","sign_citation":"https://pith.science/pith/BR5ILJPE7DL32WB55O6FOC32BV/action/citation_signature","submit_replication":"https://pith.science/pith/BR5ILJPE7DL32WB55O6FOC32BV/action/replication_record"}},"created_at":"2026-07-05T05:19:56.003073+00:00","updated_at":"2026-07-05T05:19:56.003073+00:00"}