{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:AFDZ7ZHIYYU7BAZK7TOGPWJOA5","short_pith_number":"pith:AFDZ7ZHI","schema_version":"1.0","canonical_sha256":"01479fe4e8c629f0832afcdc67d92e07712b6621bcf6d5e2ebf09c77ac2fb405","source":{"kind":"arxiv","id":"2405.13326","version":3},"attestation_state":"computed","paper":{"title":"Mosaic-IT: Cost-Free Compositional Data Synthesis for Instruction Tuning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Chenguang Wang, Fuxiao Liu, Hongyu Zhao, Ming Li, Pei Chen, Tianyi Zhou, Yijun Liang, Yupeng Hou","submitted_at":"2024-05-22T04:08:20Z","abstract_excerpt":"Finetuning large language models with a variety of instruction-response pairs has enhanced their capability to understand and follow instructions. Current instruction tuning primarily relies on teacher models or human intervention to generate and refine the instructions and responses for training, which are costly, non-sustainable, and may lack diversity. In this paper, we introduce Mosaic Instruction Tuning (Mosaic-IT), a human/model-free compositional data synthesis method that can efficiently create rich and diverse augmentations from existing instruction tuning data to enhance the LLMs. Mo"},"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":"2405.13326","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-05-22T04:08:20Z","cross_cats_sorted":[],"title_canon_sha256":"0bf0edde62acd0a80a3e68d663ebb82b4a75c06bed390a55501b8a00d297db32","abstract_canon_sha256":"e3c05c4908db4075cb850ac14651f666de7676507a1494ea16d0237e97346b36"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:16:02.256488Z","signature_b64":"klDjURVLJ/j0nXvmTy839wNuIrM6WjP6njCCAYmI4Sc5PMARLy0I/zDOuvAbJFfAKwol1uX7W/mDl8q7wz9qCA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"01479fe4e8c629f0832afcdc67d92e07712b6621bcf6d5e2ebf09c77ac2fb405","last_reissued_at":"2026-07-05T11:16:02.256002Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:16:02.256002Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Mosaic-IT: Cost-Free Compositional Data Synthesis for Instruction Tuning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Chenguang Wang, Fuxiao Liu, Hongyu Zhao, Ming Li, Pei Chen, Tianyi Zhou, Yijun Liang, Yupeng Hou","submitted_at":"2024-05-22T04:08:20Z","abstract_excerpt":"Finetuning large language models with a variety of instruction-response pairs has enhanced their capability to understand and follow instructions. Current instruction tuning primarily relies on teacher models or human intervention to generate and refine the instructions and responses for training, which are costly, non-sustainable, and may lack diversity. In this paper, we introduce Mosaic Instruction Tuning (Mosaic-IT), a human/model-free compositional data synthesis method that can efficiently create rich and diverse augmentations from existing instruction tuning data to enhance the LLMs. Mo"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2405.13326","kind":"arxiv","version":3},"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/2405.13326/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":"2405.13326","created_at":"2026-07-05T11:16:02.256058+00:00"},{"alias_kind":"arxiv_version","alias_value":"2405.13326v3","created_at":"2026-07-05T11:16:02.256058+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2405.13326","created_at":"2026-07-05T11:16:02.256058+00:00"},{"alias_kind":"pith_short_12","alias_value":"AFDZ7ZHIYYU7","created_at":"2026-07-05T11:16:02.256058+00:00"},{"alias_kind":"pith_short_16","alias_value":"AFDZ7ZHIYYU7BAZK","created_at":"2026-07-05T11:16:02.256058+00:00"},{"alias_kind":"pith_short_8","alias_value":"AFDZ7ZHI","created_at":"2026-07-05T11:16:02.256058+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2504.21850","citing_title":"Visual Compositional Tuning","ref_index":11,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/AFDZ7ZHIYYU7BAZK7TOGPWJOA5","json":"https://pith.science/pith/AFDZ7ZHIYYU7BAZK7TOGPWJOA5.json","graph_json":"https://pith.science/api/pith-number/AFDZ7ZHIYYU7BAZK7TOGPWJOA5/graph.json","events_json":"https://pith.science/api/pith-number/AFDZ7ZHIYYU7BAZK7TOGPWJOA5/events.json","paper":"https://pith.science/paper/AFDZ7ZHI"},"agent_actions":{"view_html":"https://pith.science/pith/AFDZ7ZHIYYU7BAZK7TOGPWJOA5","download_json":"https://pith.science/pith/AFDZ7ZHIYYU7BAZK7TOGPWJOA5.json","view_paper":"https://pith.science/paper/AFDZ7ZHI","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2405.13326&json=true","fetch_graph":"https://pith.science/api/pith-number/AFDZ7ZHIYYU7BAZK7TOGPWJOA5/graph.json","fetch_events":"https://pith.science/api/pith-number/AFDZ7ZHIYYU7BAZK7TOGPWJOA5/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/AFDZ7ZHIYYU7BAZK7TOGPWJOA5/action/timestamp_anchor","attest_storage":"https://pith.science/pith/AFDZ7ZHIYYU7BAZK7TOGPWJOA5/action/storage_attestation","attest_author":"https://pith.science/pith/AFDZ7ZHIYYU7BAZK7TOGPWJOA5/action/author_attestation","sign_citation":"https://pith.science/pith/AFDZ7ZHIYYU7BAZK7TOGPWJOA5/action/citation_signature","submit_replication":"https://pith.science/pith/AFDZ7ZHIYYU7BAZK7TOGPWJOA5/action/replication_record"}},"created_at":"2026-07-05T11:16:02.256058+00:00","updated_at":"2026-07-05T11:16:02.256058+00:00"}