{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:IVYAGBOXVRWUDMGKTLKTIFN2FP","short_pith_number":"pith:IVYAGBOX","schema_version":"1.0","canonical_sha256":"45700305d7ac6d41b0ca9ad53415ba2bda4a3415a0539d8a7de794d923113fb2","source":{"kind":"arxiv","id":"2307.13269","version":3},"attestation_state":"computed","paper":{"title":"LoraHub: Efficient Cross-Task Generalization via Dynamic LoRA Composition","license":"http://creativecommons.org/licenses/by-sa/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Bill Yuchen Lin, Chao Du, Chengsong Huang, Min Lin, Qian Liu, Tianyu Pang","submitted_at":"2023-07-25T05:39:21Z","abstract_excerpt":"Low-rank adaptations (LoRA) are often employed to fine-tune large language models (LLMs) for new tasks. This paper investigates LoRA composability for cross-task generalization and introduces LoraHub, a simple framework devised for the purposive assembly of LoRA modules trained on diverse given tasks, with the objective of achieving adaptable performance on unseen tasks. With just a few examples from a new task, LoraHub can fluidly combine multiple LoRA modules, eliminating the need for human expertise and assumptions. Notably, the composition requires neither additional model parameters nor g"},"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":"2307.13269","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.CL","submitted_at":"2023-07-25T05:39:21Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"18c5254be86cc9e85aecd0c32e742cedfe7fe483a9ec0f10e1aaec41541c235a","abstract_canon_sha256":"5cbee71fb1672f50f3631e46728473148605cc97dc8e078727328a59891b7beb"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:56:13.305101Z","signature_b64":"kASMhFhnQ6XlQlg1H2TfdWlGqFOWt686OlA6lnDVJ4FwfhkwchJUrzmB2P+eDzMq6yoesdzr9sAEq5s7Rjb5BQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"45700305d7ac6d41b0ca9ad53415ba2bda4a3415a0539d8a7de794d923113fb2","last_reissued_at":"2026-07-05T08:56:13.304681Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:56:13.304681Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"LoraHub: Efficient Cross-Task Generalization via Dynamic LoRA Composition","license":"http://creativecommons.org/licenses/by-sa/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Bill Yuchen Lin, Chao Du, Chengsong Huang, Min Lin, Qian Liu, Tianyu Pang","submitted_at":"2023-07-25T05:39:21Z","abstract_excerpt":"Low-rank adaptations (LoRA) are often employed to fine-tune large language models (LLMs) for new tasks. This paper investigates LoRA composability for cross-task generalization and introduces LoraHub, a simple framework devised for the purposive assembly of LoRA modules trained on diverse given tasks, with the objective of achieving adaptable performance on unseen tasks. With just a few examples from a new task, LoraHub can fluidly combine multiple LoRA modules, eliminating the need for human expertise and assumptions. Notably, the composition requires neither additional model parameters nor g"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2307.13269","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/2307.13269/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":"2307.13269","created_at":"2026-07-05T08:56:13.304738+00:00"},{"alias_kind":"arxiv_version","alias_value":"2307.13269v3","created_at":"2026-07-05T08:56:13.304738+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2307.13269","created_at":"2026-07-05T08:56:13.304738+00:00"},{"alias_kind":"pith_short_12","alias_value":"IVYAGBOXVRWU","created_at":"2026-07-05T08:56:13.304738+00:00"},{"alias_kind":"pith_short_16","alias_value":"IVYAGBOXVRWUDMGK","created_at":"2026-07-05T08:56:13.304738+00:00"},{"alias_kind":"pith_short_8","alias_value":"IVYAGBOX","created_at":"2026-07-05T08:56:13.304738+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":27,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.19172","citing_title":"User as Engram: Internalizing Per-User Memory as Local Parametric Edits","ref_index":17,"is_internal_anchor":false},{"citing_arxiv_id":"2606.19079","citing_title":"ARIADNE: Agnostic Routing for Inference-time Adapter DyNamic sElection","ref_index":20,"is_internal_anchor":false},{"citing_arxiv_id":"2606.19549","citing_title":"Predicting Mergeability of Parameter-Efficient Fine-Tuning Updates","ref_index":37,"is_internal_anchor":false},{"citing_arxiv_id":"2606.11712","citing_title":"Substrate Asymmetry in User-Side Memory: A Diagnostic Framework","ref_index":49,"is_internal_anchor":false},{"citing_arxiv_id":"2606.05889","citing_title":"GLASS: GRPO-Trained LoRA for Acoustic Style Steering in Zero-Shot Text-to-Speech","ref_index":16,"is_internal_anchor":false},{"citing_arxiv_id":"2606.06492","citing_title":"Code2LoRA: Hypernetwork-Generated Adapters for Code Language Models under Software Evolution","ref_index":14,"is_internal_anchor":false},{"citing_arxiv_id":"2606.03979","citing_title":"Language Models Need Sleep: Learning to Self-Modify and Consolidate Memories","ref_index":57,"is_internal_anchor":false},{"citing_arxiv_id":"2606.03723","citing_title":"Compress then Merge: From Multiple LoRAs into One Low-Rank Adapter","ref_index":5,"is_internal_anchor":false},{"citing_arxiv_id":"2606.01858","citing_title":"Polaris: Scaling Up Instruction-Guided Image Generation Towards Millions of Personalized Style Needs","ref_index":78,"is_internal_anchor":false},{"citing_arxiv_id":"2605.14084","citing_title":"CRANE: Constrained Reasoning Injection for Code Agents via Nullspace Editing","ref_index":3,"is_internal_anchor":false},{"citing_arxiv_id":"2605.29569","citing_title":"LoRA-Key: User-Centric LoRA Watermarking for Text-to-Image Diffusion Models","ref_index":21,"is_internal_anchor":false},{"citing_arxiv_id":"2605.29498","citing_title":"Mask the Target: A Plug-and-Play Regularizer Against LoRA Forgetting","ref_index":23,"is_internal_anchor":false},{"citing_arxiv_id":"2605.19042","citing_title":"Interference-Aware Multi-Task Unlearning","ref_index":26,"is_internal_anchor":false},{"citing_arxiv_id":"2605.16727","citing_title":"PopuLoRA: Co-Evolving LLM Populations for Reasoning Self-Play","ref_index":57,"is_internal_anchor":false},{"citing_arxiv_id":"2605.15207","citing_title":"TeamTR: Trust-Region Fine-Tuning for Multi-Agent LLM Coordination","ref_index":67,"is_internal_anchor":false},{"citing_arxiv_id":"2507.00029","citing_title":"LoRA-Mixer: Coordinate Modular LoRA Experts Through Serial Attention Routing","ref_index":15,"is_internal_anchor":false},{"citing_arxiv_id":"2310.01728","citing_title":"Time-LLM: Time Series Forecasting by Reprogramming Large Language Models","ref_index":33,"is_internal_anchor":false},{"citing_arxiv_id":"2605.14055","citing_title":"PEML: Parameter-efficient Multi-Task Learning with Optimized Continuous Prompts","ref_index":5,"is_internal_anchor":false},{"citing_arxiv_id":"2605.14084","citing_title":"CRANE: Constrained Reasoning Injection for Code Agents via Nullspace Editing","ref_index":3,"is_internal_anchor":false},{"citing_arxiv_id":"2403.14608","citing_title":"Parameter-Efficient Fine-Tuning for Large Models: A Comprehensive Survey","ref_index":92,"is_internal_anchor":false},{"citing_arxiv_id":"2605.12345","citing_title":"Output Composability of QLoRA PEFT Modules for Plug-and-Play Attribute-Controlled Text Generation","ref_index":11,"is_internal_anchor":false},{"citing_arxiv_id":"2604.23750","citing_title":"The Override Gap: A Magnitude Account of Knowledge Conflict Failure in Hypernetwork-Based Instant LLM Adaptation","ref_index":15,"is_internal_anchor":false},{"citing_arxiv_id":"2604.23750","citing_title":"The Override Gap: A Magnitude Account of Knowledge Conflict Failure in Hypernetwork-Based Instant LLM Adaptation","ref_index":15,"is_internal_anchor":false},{"citing_arxiv_id":"2604.21905","citing_title":"Low-Rank Adaptation Redux for Large Models","ref_index":72,"is_internal_anchor":false},{"citing_arxiv_id":"2605.04079","citing_title":"Efficient Handwriting-Based Alzheimer,s Disease Diagnosis Using a Low-Rank Mixture of Experts Deep Learning Framework","ref_index":30,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/IVYAGBOXVRWUDMGKTLKTIFN2FP","json":"https://pith.science/pith/IVYAGBOXVRWUDMGKTLKTIFN2FP.json","graph_json":"https://pith.science/api/pith-number/IVYAGBOXVRWUDMGKTLKTIFN2FP/graph.json","events_json":"https://pith.science/api/pith-number/IVYAGBOXVRWUDMGKTLKTIFN2FP/events.json","paper":"https://pith.science/paper/IVYAGBOX"},"agent_actions":{"view_html":"https://pith.science/pith/IVYAGBOXVRWUDMGKTLKTIFN2FP","download_json":"https://pith.science/pith/IVYAGBOXVRWUDMGKTLKTIFN2FP.json","view_paper":"https://pith.science/paper/IVYAGBOX","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2307.13269&json=true","fetch_graph":"https://pith.science/api/pith-number/IVYAGBOXVRWUDMGKTLKTIFN2FP/graph.json","fetch_events":"https://pith.science/api/pith-number/IVYAGBOXVRWUDMGKTLKTIFN2FP/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/IVYAGBOXVRWUDMGKTLKTIFN2FP/action/timestamp_anchor","attest_storage":"https://pith.science/pith/IVYAGBOXVRWUDMGKTLKTIFN2FP/action/storage_attestation","attest_author":"https://pith.science/pith/IVYAGBOXVRWUDMGKTLKTIFN2FP/action/author_attestation","sign_citation":"https://pith.science/pith/IVYAGBOXVRWUDMGKTLKTIFN2FP/action/citation_signature","submit_replication":"https://pith.science/pith/IVYAGBOXVRWUDMGKTLKTIFN2FP/action/replication_record"}},"created_at":"2026-07-05T08:56:13.304738+00:00","updated_at":"2026-07-05T08:56:13.304738+00:00"}