{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:W3YOEYWLJ5MFHWIXPBW4D7NQVL","short_pith_number":"pith:W3YOEYWL","schema_version":"1.0","canonical_sha256":"b6f0e262cb4f5853d917786dc1fdb0aac6ad96305020b05917ab65dceeea70d9","source":{"kind":"arxiv","id":"2504.15432","version":1},"attestation_state":"computed","paper":{"title":"Feeding LLM Annotations to BERT Classifiers at Your Own Risk","license":"http://creativecommons.org/licenses/by-sa/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Kazimier Smith, Yucheng Lu","submitted_at":"2025-04-21T20:54:55Z","abstract_excerpt":"Using LLM-generated labels to fine-tune smaller encoder-only models for text classification has gained popularity in various settings. While this approach may be justified in simple and low-stakes applications, we conduct empirical analysis to demonstrate how the perennial curse of training on synthetic data manifests itself in this specific setup. Compared to models trained on gold labels, we observe not only the expected performance degradation in accuracy and F1 score, but also increased instability across training runs and premature performance plateaus. These findings cast doubts on the r"},"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":"2504.15432","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.CL","submitted_at":"2025-04-21T20:54:55Z","cross_cats_sorted":[],"title_canon_sha256":"e83857f360eae6d97859d6f8ed897114304a4a6a000062007ba90b076a934ca1","abstract_canon_sha256":"5389962595353c52210bd954993946c650d530a508787c3fe8e97f42a4f1dc40"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:52:24.525843Z","signature_b64":"SVNlK5T8IVDH7yd/G0VHVVSdvoAWYyDbpJTILj+KzQCZ6m1fsSUJdmswDKgAwuWG8hJ+wZoHtUSOyaUBbFbHBw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"b6f0e262cb4f5853d917786dc1fdb0aac6ad96305020b05917ab65dceeea70d9","last_reissued_at":"2026-07-05T10:52:24.525324Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:52:24.525324Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Feeding LLM Annotations to BERT Classifiers at Your Own Risk","license":"http://creativecommons.org/licenses/by-sa/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Kazimier Smith, Yucheng Lu","submitted_at":"2025-04-21T20:54:55Z","abstract_excerpt":"Using LLM-generated labels to fine-tune smaller encoder-only models for text classification has gained popularity in various settings. While this approach may be justified in simple and low-stakes applications, we conduct empirical analysis to demonstrate how the perennial curse of training on synthetic data manifests itself in this specific setup. Compared to models trained on gold labels, we observe not only the expected performance degradation in accuracy and F1 score, but also increased instability across training runs and premature performance plateaus. These findings cast doubts on the r"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2504.15432","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/2504.15432/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":"2504.15432","created_at":"2026-07-05T10:52:24.525402+00:00"},{"alias_kind":"arxiv_version","alias_value":"2504.15432v1","created_at":"2026-07-05T10:52:24.525402+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2504.15432","created_at":"2026-07-05T10:52:24.525402+00:00"},{"alias_kind":"pith_short_12","alias_value":"W3YOEYWLJ5MF","created_at":"2026-07-05T10:52:24.525402+00:00"},{"alias_kind":"pith_short_16","alias_value":"W3YOEYWLJ5MFHWIX","created_at":"2026-07-05T10:52:24.525402+00:00"},{"alias_kind":"pith_short_8","alias_value":"W3YOEYWL","created_at":"2026-07-05T10:52:24.525402+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/W3YOEYWLJ5MFHWIXPBW4D7NQVL","json":"https://pith.science/pith/W3YOEYWLJ5MFHWIXPBW4D7NQVL.json","graph_json":"https://pith.science/api/pith-number/W3YOEYWLJ5MFHWIXPBW4D7NQVL/graph.json","events_json":"https://pith.science/api/pith-number/W3YOEYWLJ5MFHWIXPBW4D7NQVL/events.json","paper":"https://pith.science/paper/W3YOEYWL"},"agent_actions":{"view_html":"https://pith.science/pith/W3YOEYWLJ5MFHWIXPBW4D7NQVL","download_json":"https://pith.science/pith/W3YOEYWLJ5MFHWIXPBW4D7NQVL.json","view_paper":"https://pith.science/paper/W3YOEYWL","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2504.15432&json=true","fetch_graph":"https://pith.science/api/pith-number/W3YOEYWLJ5MFHWIXPBW4D7NQVL/graph.json","fetch_events":"https://pith.science/api/pith-number/W3YOEYWLJ5MFHWIXPBW4D7NQVL/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/W3YOEYWLJ5MFHWIXPBW4D7NQVL/action/timestamp_anchor","attest_storage":"https://pith.science/pith/W3YOEYWLJ5MFHWIXPBW4D7NQVL/action/storage_attestation","attest_author":"https://pith.science/pith/W3YOEYWLJ5MFHWIXPBW4D7NQVL/action/author_attestation","sign_citation":"https://pith.science/pith/W3YOEYWLJ5MFHWIXPBW4D7NQVL/action/citation_signature","submit_replication":"https://pith.science/pith/W3YOEYWLJ5MFHWIXPBW4D7NQVL/action/replication_record"}},"created_at":"2026-07-05T10:52:24.525402+00:00","updated_at":"2026-07-05T10:52:24.525402+00:00"}