{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:QQ2EMJDUAZDBNKFH3MCF63MRAL","short_pith_number":"pith:QQ2EMJDU","schema_version":"1.0","canonical_sha256":"8434462474064616a8a7db045f6d9102cb609212e130200e10dcd4de2921129c","source":{"kind":"arxiv","id":"2506.19262","version":2},"attestation_state":"computed","paper":{"title":"What Matters in LLM-generated Data: Diversity and Its Effect on Model Fine-Tuning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Haotong Wei, Huazhen Zhong, Liang Chen, Minghao Liu, Qunshu Lin, Xiaolong Sun, Yuchang Zhu, Zibin Zheng, Zixuan Yu","submitted_at":"2025-06-24T02:44:58Z","abstract_excerpt":"With the remarkable generative capabilities of large language models (LLMs), using LLM-generated data to train downstream models has emerged as a promising approach to mitigate data scarcity in specific domains and reduce time-consuming annotations. However, recent studies have highlighted a critical issue: iterative training on self-generated data results in model collapse, where model performance degrades over time. Despite extensive research on the implications of LLM-generated data, these works often neglect the importance of data diversity, a key factor in data quality. In this work, we a"},"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":"2506.19262","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2025-06-24T02:44:58Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"832d6013aad3935696d57f0484f51bb92a45e4fe5e0d6ae6abc1f8e7780ea488","abstract_canon_sha256":"35737d9b7a247a1a7040de4dc5f4196dc7dea1a92a91296fc6fa546e480706c7"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:26:53.871462Z","signature_b64":"PWGLk9ct8c1lZnLZh4MO3k6w9/ihxQuDhkalrw+ooDGM3iJ1fndSaprFchVfagKuOiELzYHGDhREUOzRtsFVDQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"8434462474064616a8a7db045f6d9102cb609212e130200e10dcd4de2921129c","last_reissued_at":"2026-07-05T11:26:53.870639Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:26:53.870639Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"What Matters in LLM-generated Data: Diversity and Its Effect on Model Fine-Tuning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Haotong Wei, Huazhen Zhong, Liang Chen, Minghao Liu, Qunshu Lin, Xiaolong Sun, Yuchang Zhu, Zibin Zheng, Zixuan Yu","submitted_at":"2025-06-24T02:44:58Z","abstract_excerpt":"With the remarkable generative capabilities of large language models (LLMs), using LLM-generated data to train downstream models has emerged as a promising approach to mitigate data scarcity in specific domains and reduce time-consuming annotations. However, recent studies have highlighted a critical issue: iterative training on self-generated data results in model collapse, where model performance degrades over time. Despite extensive research on the implications of LLM-generated data, these works often neglect the importance of data diversity, a key factor in data quality. In this work, we a"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2506.19262","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/2506.19262/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":"2506.19262","created_at":"2026-07-05T11:26:53.870743+00:00"},{"alias_kind":"arxiv_version","alias_value":"2506.19262v2","created_at":"2026-07-05T11:26:53.870743+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2506.19262","created_at":"2026-07-05T11:26:53.870743+00:00"},{"alias_kind":"pith_short_12","alias_value":"QQ2EMJDUAZDB","created_at":"2026-07-05T11:26:53.870743+00:00"},{"alias_kind":"pith_short_16","alias_value":"QQ2EMJDUAZDBNKFH","created_at":"2026-07-05T11:26:53.870743+00:00"},{"alias_kind":"pith_short_8","alias_value":"QQ2EMJDU","created_at":"2026-07-05T11:26:53.870743+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2512.15011","citing_title":"Epistemic diversity across language models mitigates knowledge collapse","ref_index":56,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/QQ2EMJDUAZDBNKFH3MCF63MRAL","json":"https://pith.science/pith/QQ2EMJDUAZDBNKFH3MCF63MRAL.json","graph_json":"https://pith.science/api/pith-number/QQ2EMJDUAZDBNKFH3MCF63MRAL/graph.json","events_json":"https://pith.science/api/pith-number/QQ2EMJDUAZDBNKFH3MCF63MRAL/events.json","paper":"https://pith.science/paper/QQ2EMJDU"},"agent_actions":{"view_html":"https://pith.science/pith/QQ2EMJDUAZDBNKFH3MCF63MRAL","download_json":"https://pith.science/pith/QQ2EMJDUAZDBNKFH3MCF63MRAL.json","view_paper":"https://pith.science/paper/QQ2EMJDU","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2506.19262&json=true","fetch_graph":"https://pith.science/api/pith-number/QQ2EMJDUAZDBNKFH3MCF63MRAL/graph.json","fetch_events":"https://pith.science/api/pith-number/QQ2EMJDUAZDBNKFH3MCF63MRAL/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/QQ2EMJDUAZDBNKFH3MCF63MRAL/action/timestamp_anchor","attest_storage":"https://pith.science/pith/QQ2EMJDUAZDBNKFH3MCF63MRAL/action/storage_attestation","attest_author":"https://pith.science/pith/QQ2EMJDUAZDBNKFH3MCF63MRAL/action/author_attestation","sign_citation":"https://pith.science/pith/QQ2EMJDUAZDBNKFH3MCF63MRAL/action/citation_signature","submit_replication":"https://pith.science/pith/QQ2EMJDUAZDBNKFH3MCF63MRAL/action/replication_record"}},"created_at":"2026-07-05T11:26:53.870743+00:00","updated_at":"2026-07-05T11:26:53.870743+00:00"}