{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:Q4POY7AO3ZCKHKZWFF4GUNDYS2","short_pith_number":"pith:Q4POY7AO","schema_version":"1.0","canonical_sha256":"871eec7c0ede44a3ab3629786a347896a33bf781af611ed6f1a32e951ae426a5","source":{"kind":"arxiv","id":"2403.11124","version":2},"attestation_state":"computed","paper":{"title":"Scaling Data Diversity for Fine-Tuning Language Models in Human Alignment","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Bowen Yu, Feifan Song, Fei Huang, Haiyang Yu, Hao Lang, Houfeng Wang, Yongbin Li","submitted_at":"2024-03-17T07:08:55Z","abstract_excerpt":"Alignment with human preference prevents large language models (LLMs) from generating misleading or toxic content while requiring high-cost human feedback. Assuming resources of human annotation are limited, there are two different ways of allocating considered: more diverse PROMPTS or more diverse RESPONSES to be labeled. Nonetheless, a straightforward comparison between their impact is absent. In this work, we first control the diversity of both sides according to the number of samples for fine-tuning, which can directly reflect their influence. We find that instead of numerous prompts, more"},"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":"2403.11124","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-03-17T07:08:55Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"928a1d2765caa0ffa174805d3568af19498c082648a066ba7911bf64cb220ac1","abstract_canon_sha256":"dff014d54f98b822be50473e75d9ad83b3e987e77110c942a361d23d53822b32"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:02:25.212116Z","signature_b64":"lcCiqkCqn0lyTLHM3qRO/6ftbFWMFv50rtpORUoA23eZHioF8EA2fItGGm4O8lJnrTyWphx2S6BD97c+bncsCg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"871eec7c0ede44a3ab3629786a347896a33bf781af611ed6f1a32e951ae426a5","last_reissued_at":"2026-07-05T08:02:25.211616Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:02:25.211616Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Scaling Data Diversity for Fine-Tuning Language Models in Human Alignment","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Bowen Yu, Feifan Song, Fei Huang, Haiyang Yu, Hao Lang, Houfeng Wang, Yongbin Li","submitted_at":"2024-03-17T07:08:55Z","abstract_excerpt":"Alignment with human preference prevents large language models (LLMs) from generating misleading or toxic content while requiring high-cost human feedback. Assuming resources of human annotation are limited, there are two different ways of allocating considered: more diverse PROMPTS or more diverse RESPONSES to be labeled. Nonetheless, a straightforward comparison between their impact is absent. In this work, we first control the diversity of both sides according to the number of samples for fine-tuning, which can directly reflect their influence. We find that instead of numerous prompts, more"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2403.11124","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/2403.11124/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":"2403.11124","created_at":"2026-07-05T08:02:25.211676+00:00"},{"alias_kind":"arxiv_version","alias_value":"2403.11124v2","created_at":"2026-07-05T08:02:25.211676+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2403.11124","created_at":"2026-07-05T08:02:25.211676+00:00"},{"alias_kind":"pith_short_12","alias_value":"Q4POY7AO3ZCK","created_at":"2026-07-05T08:02:25.211676+00:00"},{"alias_kind":"pith_short_16","alias_value":"Q4POY7AO3ZCKHKZW","created_at":"2026-07-05T08:02:25.211676+00:00"},{"alias_kind":"pith_short_8","alias_value":"Q4POY7AO","created_at":"2026-07-05T08:02:25.211676+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2502.08512","citing_title":"Measuring Diversity in Synthetic Datasets","ref_index":70,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/Q4POY7AO3ZCKHKZWFF4GUNDYS2","json":"https://pith.science/pith/Q4POY7AO3ZCKHKZWFF4GUNDYS2.json","graph_json":"https://pith.science/api/pith-number/Q4POY7AO3ZCKHKZWFF4GUNDYS2/graph.json","events_json":"https://pith.science/api/pith-number/Q4POY7AO3ZCKHKZWFF4GUNDYS2/events.json","paper":"https://pith.science/paper/Q4POY7AO"},"agent_actions":{"view_html":"https://pith.science/pith/Q4POY7AO3ZCKHKZWFF4GUNDYS2","download_json":"https://pith.science/pith/Q4POY7AO3ZCKHKZWFF4GUNDYS2.json","view_paper":"https://pith.science/paper/Q4POY7AO","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2403.11124&json=true","fetch_graph":"https://pith.science/api/pith-number/Q4POY7AO3ZCKHKZWFF4GUNDYS2/graph.json","fetch_events":"https://pith.science/api/pith-number/Q4POY7AO3ZCKHKZWFF4GUNDYS2/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/Q4POY7AO3ZCKHKZWFF4GUNDYS2/action/timestamp_anchor","attest_storage":"https://pith.science/pith/Q4POY7AO3ZCKHKZWFF4GUNDYS2/action/storage_attestation","attest_author":"https://pith.science/pith/Q4POY7AO3ZCKHKZWFF4GUNDYS2/action/author_attestation","sign_citation":"https://pith.science/pith/Q4POY7AO3ZCKHKZWFF4GUNDYS2/action/citation_signature","submit_replication":"https://pith.science/pith/Q4POY7AO3ZCKHKZWFF4GUNDYS2/action/replication_record"}},"created_at":"2026-07-05T08:02:25.211676+00:00","updated_at":"2026-07-05T08:02:25.211676+00:00"}