{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:DPYMTRM2MNQTAGAMBWXED2P46A","short_pith_number":"pith:DPYMTRM2","schema_version":"1.0","canonical_sha256":"1bf0c9c59a636130180c0dae41e9fcf005678b7e27fa49f8890c21142c6edc11","source":{"kind":"arxiv","id":"2402.02446","version":3},"attestation_state":"computed","paper":{"title":"LQER: Low-Rank Quantization Error Reconstruction for LLMs","license":"http://creativecommons.org/licenses/by-sa/4.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.LG","authors_text":"Cheng Zhang, George A. Constantinides, Jianyi Cheng, Yiren Zhao","submitted_at":"2024-02-04T10:59:52Z","abstract_excerpt":"Post-training quantization of Large Language Models (LLMs) is challenging. In this work, we introduce Low-rank Quantization Error Reduction (LQER), which combines quantization and low-rank approximation to recover the model capability. LQER leverages an activation-induced scale matrix to drive the singular value distribution of quantization error towards a desirable distribution, which enables nearly-lossless W4A8 quantization on various LLMs and downstream tasks without the need for knowledge distillation, grid search, or gradient-base iterative optimization. Unlike existing methods, the comp"},"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":"2402.02446","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.LG","submitted_at":"2024-02-04T10:59:52Z","cross_cats_sorted":["cs.CL"],"title_canon_sha256":"f84fc19071b9a586822f7bd40d1aaf68b1cc218b469d2a9fedaac130b5ff0686","abstract_canon_sha256":"a7cd53f49be71a00e15b97eb1bc15bfcbe9031e468ac5231bbf5757b125677e1"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:25:00.035198Z","signature_b64":"b1NRd5WRxZelopMCTFQxp/n1wk0gzkTVNEmlf0D6oTcNeILuOBIlMRoU33rGOKetdqea7wDUyigu8lPsm7qOAQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"1bf0c9c59a636130180c0dae41e9fcf005678b7e27fa49f8890c21142c6edc11","last_reissued_at":"2026-07-05T08:25:00.034742Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:25:00.034742Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"LQER: Low-Rank Quantization Error Reconstruction for LLMs","license":"http://creativecommons.org/licenses/by-sa/4.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.LG","authors_text":"Cheng Zhang, George A. Constantinides, Jianyi Cheng, Yiren Zhao","submitted_at":"2024-02-04T10:59:52Z","abstract_excerpt":"Post-training quantization of Large Language Models (LLMs) is challenging. In this work, we introduce Low-rank Quantization Error Reduction (LQER), which combines quantization and low-rank approximation to recover the model capability. LQER leverages an activation-induced scale matrix to drive the singular value distribution of quantization error towards a desirable distribution, which enables nearly-lossless W4A8 quantization on various LLMs and downstream tasks without the need for knowledge distillation, grid search, or gradient-base iterative optimization. Unlike existing methods, the comp"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2402.02446","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/2402.02446/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":"2402.02446","created_at":"2026-07-05T08:25:00.034812+00:00"},{"alias_kind":"arxiv_version","alias_value":"2402.02446v3","created_at":"2026-07-05T08:25:00.034812+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2402.02446","created_at":"2026-07-05T08:25:00.034812+00:00"},{"alias_kind":"pith_short_12","alias_value":"DPYMTRM2MNQT","created_at":"2026-07-05T08:25:00.034812+00:00"},{"alias_kind":"pith_short_16","alias_value":"DPYMTRM2MNQTAGAM","created_at":"2026-07-05T08:25:00.034812+00:00"},{"alias_kind":"pith_short_8","alias_value":"DPYMTRM2","created_at":"2026-07-05T08:25:00.034812+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.01412","citing_title":"GPTQ-intrinsic LoRA: A Near-optimal Algorithm for Low-precision Quantization with Low-rank Adaptation","ref_index":78,"is_internal_anchor":false},{"citing_arxiv_id":"2505.12942","citing_title":"A3 : an Analytical Low-Rank Approximation Framework for Attention","ref_index":19,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/DPYMTRM2MNQTAGAMBWXED2P46A","json":"https://pith.science/pith/DPYMTRM2MNQTAGAMBWXED2P46A.json","graph_json":"https://pith.science/api/pith-number/DPYMTRM2MNQTAGAMBWXED2P46A/graph.json","events_json":"https://pith.science/api/pith-number/DPYMTRM2MNQTAGAMBWXED2P46A/events.json","paper":"https://pith.science/paper/DPYMTRM2"},"agent_actions":{"view_html":"https://pith.science/pith/DPYMTRM2MNQTAGAMBWXED2P46A","download_json":"https://pith.science/pith/DPYMTRM2MNQTAGAMBWXED2P46A.json","view_paper":"https://pith.science/paper/DPYMTRM2","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2402.02446&json=true","fetch_graph":"https://pith.science/api/pith-number/DPYMTRM2MNQTAGAMBWXED2P46A/graph.json","fetch_events":"https://pith.science/api/pith-number/DPYMTRM2MNQTAGAMBWXED2P46A/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/DPYMTRM2MNQTAGAMBWXED2P46A/action/timestamp_anchor","attest_storage":"https://pith.science/pith/DPYMTRM2MNQTAGAMBWXED2P46A/action/storage_attestation","attest_author":"https://pith.science/pith/DPYMTRM2MNQTAGAMBWXED2P46A/action/author_attestation","sign_citation":"https://pith.science/pith/DPYMTRM2MNQTAGAMBWXED2P46A/action/citation_signature","submit_replication":"https://pith.science/pith/DPYMTRM2MNQTAGAMBWXED2P46A/action/replication_record"}},"created_at":"2026-07-05T08:25:00.034812+00:00","updated_at":"2026-07-05T08:25:00.034812+00:00"}