{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:YTD2BKHM4UX3KUZ3ZFNMB7QI22","short_pith_number":"pith:YTD2BKHM","schema_version":"1.0","canonical_sha256":"c4c7a0a8ece52fb5533bc95ac0fe08d6812c818d56eeb33c995851201e9b9239","source":{"kind":"arxiv","id":"2410.10877","version":2},"attestation_state":"computed","paper":{"title":"Improving Data Efficiency via Curating LLM-Driven Rating Systems","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Ankit Parag Shah, Chen Qian, Jiaheng Wei, Jinlong Pang, Wei Wei, Yang Liu, Yaxuan Wang, Yujia Bao, Zhaowei Zhu","submitted_at":"2024-10-09T10:07:55Z","abstract_excerpt":"Instruction tuning is critical for adapting large language models (LLMs) to downstream tasks, and recent studies have demonstrated that small amounts of human-curated data can outperform larger datasets, challenging traditional data scaling laws. While LLM-based data quality rating systems offer a cost-effective alternative to human annotation, they often suffer from inaccuracies and biases, even in powerful models like GPT-4. In this work, we introduce DS2, a Diversity-aware Score curation method for Data Selection. By systematically modeling error patterns through a score transition matrix, "},"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":"2410.10877","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-10-09T10:07:55Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"3bc598cf64b4e68524fe87ded419fd91925130e094cc0a1e679e354f025c9b81","abstract_canon_sha256":"8f46ff15397471b0e2206e506fc92cd931ef8142693d545835bcf8878599f557"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:25:11.745925Z","signature_b64":"pZgM0kjElEBCBaCbDu08ODfbdeTGQ30nzLndF0osHAsVFoVn56O6psYyiveMBIZRPhQX/vclsk6oxTZgu+mXCw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"c4c7a0a8ece52fb5533bc95ac0fe08d6812c818d56eeb33c995851201e9b9239","last_reissued_at":"2026-07-05T10:25:11.745050Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:25:11.745050Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Improving Data Efficiency via Curating LLM-Driven Rating Systems","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Ankit Parag Shah, Chen Qian, Jiaheng Wei, Jinlong Pang, Wei Wei, Yang Liu, Yaxuan Wang, Yujia Bao, Zhaowei Zhu","submitted_at":"2024-10-09T10:07:55Z","abstract_excerpt":"Instruction tuning is critical for adapting large language models (LLMs) to downstream tasks, and recent studies have demonstrated that small amounts of human-curated data can outperform larger datasets, challenging traditional data scaling laws. While LLM-based data quality rating systems offer a cost-effective alternative to human annotation, they often suffer from inaccuracies and biases, even in powerful models like GPT-4. In this work, we introduce DS2, a Diversity-aware Score curation method for Data Selection. By systematically modeling error patterns through a score transition matrix, "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2410.10877","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/2410.10877/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":"2410.10877","created_at":"2026-07-05T10:25:11.745173+00:00"},{"alias_kind":"arxiv_version","alias_value":"2410.10877v2","created_at":"2026-07-05T10:25:11.745173+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2410.10877","created_at":"2026-07-05T10:25:11.745173+00:00"},{"alias_kind":"pith_short_12","alias_value":"YTD2BKHM4UX3","created_at":"2026-07-05T10:25:11.745173+00:00"},{"alias_kind":"pith_short_16","alias_value":"YTD2BKHM4UX3KUZ3","created_at":"2026-07-05T10:25:11.745173+00:00"},{"alias_kind":"pith_short_8","alias_value":"YTD2BKHM","created_at":"2026-07-05T10:25:11.745173+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.18286","citing_title":"CODEBLOCK: Learning to Supervise Code at the Right Granularity","ref_index":16,"is_internal_anchor":false},{"citing_arxiv_id":"2605.30857","citing_title":"MADS: Model-Aware Diverse Core Set Selection for Instruction Tuning","ref_index":2,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/YTD2BKHM4UX3KUZ3ZFNMB7QI22","json":"https://pith.science/pith/YTD2BKHM4UX3KUZ3ZFNMB7QI22.json","graph_json":"https://pith.science/api/pith-number/YTD2BKHM4UX3KUZ3ZFNMB7QI22/graph.json","events_json":"https://pith.science/api/pith-number/YTD2BKHM4UX3KUZ3ZFNMB7QI22/events.json","paper":"https://pith.science/paper/YTD2BKHM"},"agent_actions":{"view_html":"https://pith.science/pith/YTD2BKHM4UX3KUZ3ZFNMB7QI22","download_json":"https://pith.science/pith/YTD2BKHM4UX3KUZ3ZFNMB7QI22.json","view_paper":"https://pith.science/paper/YTD2BKHM","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2410.10877&json=true","fetch_graph":"https://pith.science/api/pith-number/YTD2BKHM4UX3KUZ3ZFNMB7QI22/graph.json","fetch_events":"https://pith.science/api/pith-number/YTD2BKHM4UX3KUZ3ZFNMB7QI22/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/YTD2BKHM4UX3KUZ3ZFNMB7QI22/action/timestamp_anchor","attest_storage":"https://pith.science/pith/YTD2BKHM4UX3KUZ3ZFNMB7QI22/action/storage_attestation","attest_author":"https://pith.science/pith/YTD2BKHM4UX3KUZ3ZFNMB7QI22/action/author_attestation","sign_citation":"https://pith.science/pith/YTD2BKHM4UX3KUZ3ZFNMB7QI22/action/citation_signature","submit_replication":"https://pith.science/pith/YTD2BKHM4UX3KUZ3ZFNMB7QI22/action/replication_record"}},"created_at":"2026-07-05T10:25:11.745173+00:00","updated_at":"2026-07-05T10:25:11.745173+00:00"}