{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:WSLRJZE7C4R5Q7YIHRTNG6Y65G","short_pith_number":"pith:WSLRJZE7","schema_version":"1.0","canonical_sha256":"b49714e49f1723d87f083c66d37b1ee9b259868e9fcb566f69e67762be2b009d","source":{"kind":"arxiv","id":"2406.01721","version":3},"attestation_state":"computed","paper":{"title":"DuQuant: Distributing Outliers via Dual Transformation Makes Stronger Quantized LLMs","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Haobo Xu, Haokun Lin, Jingzhi Cui, Linqi Song, Linzhan Mou, Yichen Wu, Yingtao Zhang, Ying Wei, Zhenan Sun","submitted_at":"2024-06-03T18:27:44Z","abstract_excerpt":"Quantization of large language models (LLMs) faces significant challenges, particularly due to the presence of outlier activations that impede efficient low-bit representation. Traditional approaches predominantly address Normal Outliers, which are activations across all tokens with relatively large magnitudes. However, these methods struggle with smoothing Massive Outliers that display significantly larger values, which leads to significant performance degradation in low-bit quantization. In this paper, we introduce DuQuant, a novel approach that utilizes rotation and permutation transformati"},"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":"2406.01721","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-06-03T18:27:44Z","cross_cats_sorted":[],"title_canon_sha256":"535a7cebdf200ca14f68f2dcb4dec5eafcbc8e84e280a5f3fa8c55f9af487f1b","abstract_canon_sha256":"88e3f9c3a937661793f15b4ef376aa168993f13fcce941e4ff07decad109e9ed"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:29:29.887793Z","signature_b64":"dBRdQGQ34+ObXsjEdpmpdhbCJxNi3Crg7dy5NLQvhOvY62aaUQDM1/yDJbr9wEoI7mXwPvYoEuKaDT5AfQGvBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"b49714e49f1723d87f083c66d37b1ee9b259868e9fcb566f69e67762be2b009d","last_reissued_at":"2026-07-05T09:29:29.887273Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:29:29.887273Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"DuQuant: Distributing Outliers via Dual Transformation Makes Stronger Quantized LLMs","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Haobo Xu, Haokun Lin, Jingzhi Cui, Linqi Song, Linzhan Mou, Yichen Wu, Yingtao Zhang, Ying Wei, Zhenan Sun","submitted_at":"2024-06-03T18:27:44Z","abstract_excerpt":"Quantization of large language models (LLMs) faces significant challenges, particularly due to the presence of outlier activations that impede efficient low-bit representation. Traditional approaches predominantly address Normal Outliers, which are activations across all tokens with relatively large magnitudes. However, these methods struggle with smoothing Massive Outliers that display significantly larger values, which leads to significant performance degradation in low-bit quantization. In this paper, we introduce DuQuant, a novel approach that utilizes rotation and permutation transformati"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2406.01721","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/2406.01721/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":"2406.01721","created_at":"2026-07-05T09:29:29.887343+00:00"},{"alias_kind":"arxiv_version","alias_value":"2406.01721v3","created_at":"2026-07-05T09:29:29.887343+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2406.01721","created_at":"2026-07-05T09:29:29.887343+00:00"},{"alias_kind":"pith_short_12","alias_value":"WSLRJZE7C4R5","created_at":"2026-07-05T09:29:29.887343+00:00"},{"alias_kind":"pith_short_16","alias_value":"WSLRJZE7C4R5Q7YI","created_at":"2026-07-05T09:29:29.887343+00:00"},{"alias_kind":"pith_short_8","alias_value":"WSLRJZE7","created_at":"2026-07-05T09:29:29.887343+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.26587","citing_title":"SharQ: Bridging Activation Sparsity and FP4 Quantization for LLM Inference","ref_index":71,"is_internal_anchor":false},{"citing_arxiv_id":"2605.26558","citing_title":"Cassandra: Enabling Reasoning LLMs at Edge via Self-Speculative Decoding","ref_index":32,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/WSLRJZE7C4R5Q7YIHRTNG6Y65G","json":"https://pith.science/pith/WSLRJZE7C4R5Q7YIHRTNG6Y65G.json","graph_json":"https://pith.science/api/pith-number/WSLRJZE7C4R5Q7YIHRTNG6Y65G/graph.json","events_json":"https://pith.science/api/pith-number/WSLRJZE7C4R5Q7YIHRTNG6Y65G/events.json","paper":"https://pith.science/paper/WSLRJZE7"},"agent_actions":{"view_html":"https://pith.science/pith/WSLRJZE7C4R5Q7YIHRTNG6Y65G","download_json":"https://pith.science/pith/WSLRJZE7C4R5Q7YIHRTNG6Y65G.json","view_paper":"https://pith.science/paper/WSLRJZE7","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2406.01721&json=true","fetch_graph":"https://pith.science/api/pith-number/WSLRJZE7C4R5Q7YIHRTNG6Y65G/graph.json","fetch_events":"https://pith.science/api/pith-number/WSLRJZE7C4R5Q7YIHRTNG6Y65G/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/WSLRJZE7C4R5Q7YIHRTNG6Y65G/action/timestamp_anchor","attest_storage":"https://pith.science/pith/WSLRJZE7C4R5Q7YIHRTNG6Y65G/action/storage_attestation","attest_author":"https://pith.science/pith/WSLRJZE7C4R5Q7YIHRTNG6Y65G/action/author_attestation","sign_citation":"https://pith.science/pith/WSLRJZE7C4R5Q7YIHRTNG6Y65G/action/citation_signature","submit_replication":"https://pith.science/pith/WSLRJZE7C4R5Q7YIHRTNG6Y65G/action/replication_record"}},"created_at":"2026-07-05T09:29:29.887343+00:00","updated_at":"2026-07-05T09:29:29.887343+00:00"}