{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:Q2WALCE7VVGVPQEIZN42KHQTMT","short_pith_number":"pith:Q2WALCE7","schema_version":"1.0","canonical_sha256":"86ac05889fad4d57c088cb79a51e1364c85b067fc43bb5345422210a72e48afc","source":{"kind":"arxiv","id":"2503.23714","version":2},"attestation_state":"computed","paper":{"title":"Building Instruction-Tuning Datasets from Human-Written Instructions with Open-Weight Large Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Hinari Shimada, Hiroya Takamura, Kakeru Hattori, Kazuki Fujii, Koki Maeda, Koshiro Saito, Masanari Ohi, Naoaki Okazaki, Rio Yokota, Sakae Mizuki, Shigeki Ishida, Taihei Shiotani, Taishi Nakamura, Takumi Okamoto, Youmi Ma","submitted_at":"2025-03-31T04:28:38Z","abstract_excerpt":"Instruction tuning is crucial for enabling Large Language Models (LLMs) to solve real-world tasks. Prior work has shown the effectiveness of instruction-tuning data synthesized solely from LLMs, raising a fundamental question: Do we still need human-originated signals for instruction tuning? This work answers the question affirmatively: we build state-of-the-art instruction-tuning datasets sourced from human-written instructions, by simply pairing them with LLM-generated responses. LLMs fine-tuned on our datasets consistently outperform those fine-tuned on existing ones. Our data construction "},"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":"2503.23714","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-03-31T04:28:38Z","cross_cats_sorted":[],"title_canon_sha256":"f72e1b8a876d3b0f8600c778f3faded5b76e632ccb5c617398518c79666f38b2","abstract_canon_sha256":"eef06e9ea4113169aacc0c9f0dcfe8f9e684c35f718dec0b15db955768090133"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:53:45.219153Z","signature_b64":"yHIccrQjKnoyBDGzJKFH9M7Soksy0m4wowqCCKWhoSgNnGNCOtuvJKhhIeG5ClfPVMhbJkEUeCnpRa64OiQqDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"86ac05889fad4d57c088cb79a51e1364c85b067fc43bb5345422210a72e48afc","last_reissued_at":"2026-07-05T11:53:45.218647Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:53:45.218647Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Building Instruction-Tuning Datasets from Human-Written Instructions with Open-Weight Large Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Hinari Shimada, Hiroya Takamura, Kakeru Hattori, Kazuki Fujii, Koki Maeda, Koshiro Saito, Masanari Ohi, Naoaki Okazaki, Rio Yokota, Sakae Mizuki, Shigeki Ishida, Taihei Shiotani, Taishi Nakamura, Takumi Okamoto, Youmi Ma","submitted_at":"2025-03-31T04:28:38Z","abstract_excerpt":"Instruction tuning is crucial for enabling Large Language Models (LLMs) to solve real-world tasks. Prior work has shown the effectiveness of instruction-tuning data synthesized solely from LLMs, raising a fundamental question: Do we still need human-originated signals for instruction tuning? This work answers the question affirmatively: we build state-of-the-art instruction-tuning datasets sourced from human-written instructions, by simply pairing them with LLM-generated responses. LLMs fine-tuned on our datasets consistently outperform those fine-tuned on existing ones. Our data construction "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2503.23714","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/2503.23714/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":"2503.23714","created_at":"2026-07-05T11:53:45.218709+00:00"},{"alias_kind":"arxiv_version","alias_value":"2503.23714v2","created_at":"2026-07-05T11:53:45.218709+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2503.23714","created_at":"2026-07-05T11:53:45.218709+00:00"},{"alias_kind":"pith_short_12","alias_value":"Q2WALCE7VVGV","created_at":"2026-07-05T11:53:45.218709+00:00"},{"alias_kind":"pith_short_16","alias_value":"Q2WALCE7VVGVPQEI","created_at":"2026-07-05T11:53:45.218709+00:00"},{"alias_kind":"pith_short_8","alias_value":"Q2WALCE7","created_at":"2026-07-05T11:53:45.218709+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2505.16418","citing_title":"Exploring the Relationship Between Diversity and Quality in Ad Text Generation","ref_index":18,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/Q2WALCE7VVGVPQEIZN42KHQTMT","json":"https://pith.science/pith/Q2WALCE7VVGVPQEIZN42KHQTMT.json","graph_json":"https://pith.science/api/pith-number/Q2WALCE7VVGVPQEIZN42KHQTMT/graph.json","events_json":"https://pith.science/api/pith-number/Q2WALCE7VVGVPQEIZN42KHQTMT/events.json","paper":"https://pith.science/paper/Q2WALCE7"},"agent_actions":{"view_html":"https://pith.science/pith/Q2WALCE7VVGVPQEIZN42KHQTMT","download_json":"https://pith.science/pith/Q2WALCE7VVGVPQEIZN42KHQTMT.json","view_paper":"https://pith.science/paper/Q2WALCE7","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2503.23714&json=true","fetch_graph":"https://pith.science/api/pith-number/Q2WALCE7VVGVPQEIZN42KHQTMT/graph.json","fetch_events":"https://pith.science/api/pith-number/Q2WALCE7VVGVPQEIZN42KHQTMT/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/Q2WALCE7VVGVPQEIZN42KHQTMT/action/timestamp_anchor","attest_storage":"https://pith.science/pith/Q2WALCE7VVGVPQEIZN42KHQTMT/action/storage_attestation","attest_author":"https://pith.science/pith/Q2WALCE7VVGVPQEIZN42KHQTMT/action/author_attestation","sign_citation":"https://pith.science/pith/Q2WALCE7VVGVPQEIZN42KHQTMT/action/citation_signature","submit_replication":"https://pith.science/pith/Q2WALCE7VVGVPQEIZN42KHQTMT/action/replication_record"}},"created_at":"2026-07-05T11:53:45.218709+00:00","updated_at":"2026-07-05T11:53:45.218709+00:00"}