{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:JEEPV7J6USTKJSRQY4QGGUU52C","short_pith_number":"pith:JEEPV7J6","schema_version":"1.0","canonical_sha256":"4908fafd3ea4a6a4ca30c72063529dd09591f21ab4e047072710c6a19c6242ea","source":{"kind":"arxiv","id":"2501.09368","version":4},"attestation_state":"computed","paper":{"title":"Aligning Instruction Tuning with Pre-training","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Chujie Zheng, Ge Zhang, Guoyin Wang, Jiaheng Liu, Jiajun Zhang, Lei Ma, Tianyu Zheng, Wenhao Huang, Wenqiang Zu, Xiang Yue, Xingrun Xing, Xingwei Qu, Xinrun Du, Yiming Liang, Zhaoxiang Zhang","submitted_at":"2025-01-16T08:27:40Z","abstract_excerpt":"Instruction tuning enhances large language models (LLMs) to follow human instructions across diverse tasks, relying on high-quality datasets to guide behavior. However, these datasets, whether manually curated or synthetically generated, are often narrowly focused and misaligned with the broad distributions captured during pre-training, limiting LLM generalization and effective use of pre-trained knowledge. We propose Aligning Instruction Tuning with Pre-training (AITP), a method that bridges this gap by identifying coverage shortfalls in instruction-tuning datasets and rewriting underrepresen"},"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":"2501.09368","kind":"arxiv","version":4},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2025-01-16T08:27:40Z","cross_cats_sorted":[],"title_canon_sha256":"eda60d2cad23d13f52542a6c5ddd9e34d1eccf0762bafecb651fce575b0708bf","abstract_canon_sha256":"8521e1d6d81e577f1c4fc1239508b81df6c90caa57b7cf41658dce198eadc778"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:51:21.027374Z","signature_b64":"plCX6ak0tLZFpIJH+PE40NytJLOeI8D5SP0atH6eV1apY3qsp/Rgey8cskw4cj2Q03Gul2E4Wks1hMz4PPy7Cg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"4908fafd3ea4a6a4ca30c72063529dd09591f21ab4e047072710c6a19c6242ea","last_reissued_at":"2026-07-05T11:51:21.026885Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:51:21.026885Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Aligning Instruction Tuning with Pre-training","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Chujie Zheng, Ge Zhang, Guoyin Wang, Jiaheng Liu, Jiajun Zhang, Lei Ma, Tianyu Zheng, Wenhao Huang, Wenqiang Zu, Xiang Yue, Xingrun Xing, Xingwei Qu, Xinrun Du, Yiming Liang, Zhaoxiang Zhang","submitted_at":"2025-01-16T08:27:40Z","abstract_excerpt":"Instruction tuning enhances large language models (LLMs) to follow human instructions across diverse tasks, relying on high-quality datasets to guide behavior. However, these datasets, whether manually curated or synthetically generated, are often narrowly focused and misaligned with the broad distributions captured during pre-training, limiting LLM generalization and effective use of pre-trained knowledge. We propose Aligning Instruction Tuning with Pre-training (AITP), a method that bridges this gap by identifying coverage shortfalls in instruction-tuning datasets and rewriting underrepresen"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2501.09368","kind":"arxiv","version":4},"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/2501.09368/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":"2501.09368","created_at":"2026-07-05T11:51:21.026940+00:00"},{"alias_kind":"arxiv_version","alias_value":"2501.09368v4","created_at":"2026-07-05T11:51:21.026940+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2501.09368","created_at":"2026-07-05T11:51:21.026940+00:00"},{"alias_kind":"pith_short_12","alias_value":"JEEPV7J6USTK","created_at":"2026-07-05T11:51:21.026940+00:00"},{"alias_kind":"pith_short_16","alias_value":"JEEPV7J6USTKJSRQ","created_at":"2026-07-05T11:51:21.026940+00:00"},{"alias_kind":"pith_short_8","alias_value":"JEEPV7J6","created_at":"2026-07-05T11:51:21.026940+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2506.04288","citing_title":"Backbone Augmented Training for Adaptations","ref_index":17,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/JEEPV7J6USTKJSRQY4QGGUU52C","json":"https://pith.science/pith/JEEPV7J6USTKJSRQY4QGGUU52C.json","graph_json":"https://pith.science/api/pith-number/JEEPV7J6USTKJSRQY4QGGUU52C/graph.json","events_json":"https://pith.science/api/pith-number/JEEPV7J6USTKJSRQY4QGGUU52C/events.json","paper":"https://pith.science/paper/JEEPV7J6"},"agent_actions":{"view_html":"https://pith.science/pith/JEEPV7J6USTKJSRQY4QGGUU52C","download_json":"https://pith.science/pith/JEEPV7J6USTKJSRQY4QGGUU52C.json","view_paper":"https://pith.science/paper/JEEPV7J6","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2501.09368&json=true","fetch_graph":"https://pith.science/api/pith-number/JEEPV7J6USTKJSRQY4QGGUU52C/graph.json","fetch_events":"https://pith.science/api/pith-number/JEEPV7J6USTKJSRQY4QGGUU52C/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/JEEPV7J6USTKJSRQY4QGGUU52C/action/timestamp_anchor","attest_storage":"https://pith.science/pith/JEEPV7J6USTKJSRQY4QGGUU52C/action/storage_attestation","attest_author":"https://pith.science/pith/JEEPV7J6USTKJSRQY4QGGUU52C/action/author_attestation","sign_citation":"https://pith.science/pith/JEEPV7J6USTKJSRQY4QGGUU52C/action/citation_signature","submit_replication":"https://pith.science/pith/JEEPV7J6USTKJSRQY4QGGUU52C/action/replication_record"}},"created_at":"2026-07-05T11:51:21.026940+00:00","updated_at":"2026-07-05T11:51:21.026940+00:00"}