{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:URQTB6NQCIHFINIQMYJYOXFXDD","short_pith_number":"pith:URQTB6NQ","schema_version":"1.0","canonical_sha256":"a46130f9b0120e5435106613875cb718c644472ea22c74d402f75ba5c84a5105","source":{"kind":"arxiv","id":"2310.08659","version":4},"attestation_state":"computed","paper":{"title":"LoftQ: LoRA-Fine-Tuning-Aware Quantization for Large Language Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"Chen Liang, Nikos Karampatziakis, Pengcheng He, Tuo Zhao, Weizhu Chen, Yifan Yu, Yixiao Li","submitted_at":"2023-10-12T18:34:08Z","abstract_excerpt":"Quantization is an indispensable technique for serving Large Language Models (LLMs) and has recently found its way into LoRA fine-tuning. In this work we focus on the scenario where quantization and LoRA fine-tuning are applied together on a pre-trained model. In such cases it is common to observe a consistent gap in the performance on downstream tasks between full fine-tuning and quantization plus LoRA fine-tuning approach. In response, we propose LoftQ (LoRA-Fine-Tuning-aware Quantization), a novel quantization framework that simultaneously quantizes an LLM and finds a proper low-rank initia"},"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":"2310.08659","kind":"arxiv","version":4},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2023-10-12T18:34:08Z","cross_cats_sorted":["cs.AI","cs.LG"],"title_canon_sha256":"de896ef0e63615a9cb84dc80d0679356728a455633518456a2033e1600f625c3","abstract_canon_sha256":"6c9ca242d6b8c9ea049d77f1fd56a26e4e484107fd8c6ec29a075176c8590ebe"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T07:17:32.866650Z","signature_b64":"cArRQyDQDlkzxnCiO4Ep6ehao6FGPmnNfGLpsIAsh/1aWDwwcc1r3uksnX8A+fEaBZ5BIeA7wGKs62b03OuSDg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"a46130f9b0120e5435106613875cb718c644472ea22c74d402f75ba5c84a5105","last_reissued_at":"2026-07-05T07:17:32.866133Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T07:17:32.866133Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"LoftQ: LoRA-Fine-Tuning-Aware Quantization for Large Language Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"Chen Liang, Nikos Karampatziakis, Pengcheng He, Tuo Zhao, Weizhu Chen, Yifan Yu, Yixiao Li","submitted_at":"2023-10-12T18:34:08Z","abstract_excerpt":"Quantization is an indispensable technique for serving Large Language Models (LLMs) and has recently found its way into LoRA fine-tuning. In this work we focus on the scenario where quantization and LoRA fine-tuning are applied together on a pre-trained model. In such cases it is common to observe a consistent gap in the performance on downstream tasks between full fine-tuning and quantization plus LoRA fine-tuning approach. In response, we propose LoftQ (LoRA-Fine-Tuning-aware Quantization), a novel quantization framework that simultaneously quantizes an LLM and finds a proper low-rank initia"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2310.08659","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/2310.08659/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":"2310.08659","created_at":"2026-07-05T07:17:32.866195+00:00"},{"alias_kind":"arxiv_version","alias_value":"2310.08659v4","created_at":"2026-07-05T07:17:32.866195+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2310.08659","created_at":"2026-07-05T07:17:32.866195+00:00"},{"alias_kind":"pith_short_12","alias_value":"URQTB6NQCIHF","created_at":"2026-07-05T07:17:32.866195+00:00"},{"alias_kind":"pith_short_16","alias_value":"URQTB6NQCIHFINIQ","created_at":"2026-07-05T07:17:32.866195+00:00"},{"alias_kind":"pith_short_8","alias_value":"URQTB6NQ","created_at":"2026-07-05T07:17:32.866195+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":13,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2607.08194","citing_title":"Dive Into the Implicit Biases of Low-rank Vision-language Alignment","ref_index":23,"is_internal_anchor":true},{"citing_arxiv_id":"2606.11244","citing_title":"SPEAR: A System for Post-Quantization Error-Adaptive Recovery Enabling Efficient Low-Bit LLM Serving","ref_index":28,"is_internal_anchor":false},{"citing_arxiv_id":"2606.01412","citing_title":"GPTQ-intrinsic LoRA: A Near-optimal Algorithm for Low-precision Quantization with Low-rank Adaptation","ref_index":41,"is_internal_anchor":false},{"citing_arxiv_id":"2605.14841","citing_title":"GPart: End-to-End Isometric Fine-Tuning via Global Parameter Partitioning","ref_index":2,"is_internal_anchor":false},{"citing_arxiv_id":"2606.00494","citing_title":"ProjQ: Project-and-Quantize for Adapter-Aware LLM Compression","ref_index":69,"is_internal_anchor":false},{"citing_arxiv_id":"2409.18169","citing_title":"Harmful Fine-tuning Attacks and Defenses for Large Language Models: A Survey","ref_index":90,"is_internal_anchor":false},{"citing_arxiv_id":"2508.06974","citing_title":"Rethinking 1-bit Optimization Leveraging Pre-trained Large Language Models","ref_index":15,"is_internal_anchor":false},{"citing_arxiv_id":"2507.00029","citing_title":"LoRA-Mixer: Coordinate Modular LoRA Experts Through Serial Attention Routing","ref_index":51,"is_internal_anchor":false},{"citing_arxiv_id":"2404.14294","citing_title":"A Survey on Efficient Inference for Large Language Models","ref_index":191,"is_internal_anchor":false},{"citing_arxiv_id":"2604.02501","citing_title":"ECG Foundation Models and Medical LLMs for Agentic Cardiovascular Intelligence at the Edge: A Review and Outlook","ref_index":95,"is_internal_anchor":false},{"citing_arxiv_id":"2403.14608","citing_title":"Parameter-Efficient Fine-Tuning for Large Models: A Comprehensive Survey","ref_index":125,"is_internal_anchor":false},{"citing_arxiv_id":"2604.27763","citing_title":"Intent2Tx: Benchmarking LLMs for Translating Natural Language Intents into Ethereum Transactions","ref_index":14,"is_internal_anchor":false},{"citing_arxiv_id":"2605.05819","citing_title":"HCInfer: An Efficient Inference System via Error Compensation for Resource-Constrained Devices","ref_index":25,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/URQTB6NQCIHFINIQMYJYOXFXDD","json":"https://pith.science/pith/URQTB6NQCIHFINIQMYJYOXFXDD.json","graph_json":"https://pith.science/api/pith-number/URQTB6NQCIHFINIQMYJYOXFXDD/graph.json","events_json":"https://pith.science/api/pith-number/URQTB6NQCIHFINIQMYJYOXFXDD/events.json","paper":"https://pith.science/paper/URQTB6NQ"},"agent_actions":{"view_html":"https://pith.science/pith/URQTB6NQCIHFINIQMYJYOXFXDD","download_json":"https://pith.science/pith/URQTB6NQCIHFINIQMYJYOXFXDD.json","view_paper":"https://pith.science/paper/URQTB6NQ","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2310.08659&json=true","fetch_graph":"https://pith.science/api/pith-number/URQTB6NQCIHFINIQMYJYOXFXDD/graph.json","fetch_events":"https://pith.science/api/pith-number/URQTB6NQCIHFINIQMYJYOXFXDD/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/URQTB6NQCIHFINIQMYJYOXFXDD/action/timestamp_anchor","attest_storage":"https://pith.science/pith/URQTB6NQCIHFINIQMYJYOXFXDD/action/storage_attestation","attest_author":"https://pith.science/pith/URQTB6NQCIHFINIQMYJYOXFXDD/action/author_attestation","sign_citation":"https://pith.science/pith/URQTB6NQCIHFINIQMYJYOXFXDD/action/citation_signature","submit_replication":"https://pith.science/pith/URQTB6NQCIHFINIQMYJYOXFXDD/action/replication_record"}},"created_at":"2026-07-05T07:17:32.866195+00:00","updated_at":"2026-07-05T07:17:32.866195+00:00"}