{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2024:5P27RQRK5X2O3D4JCX43JFPLFQ","short_pith_number":"pith:5P27RQRK","canonical_record":{"source":{"id":"2410.05265","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-10-07T17:59:35Z","cross_cats_sorted":["cs.CL"],"title_canon_sha256":"c31cfb2b4566ce2cc2e14d25a954bc6bb14856d20c5ee49215fb18c00825923a","abstract_canon_sha256":"06338fd17b50ec208c0476c5f8bbfc132cba3138b68e71832d26f3679daa7f08"},"schema_version":"1.0"},"canonical_sha256":"ebf5f8c22aedf4ed8f8915f9b495eb2c356b9336bb7f8248324a8e05c04f2504","source":{"kind":"arxiv","id":"2410.05265","version":2},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2410.05265","created_at":"2026-07-05T10:05:45Z"},{"alias_kind":"arxiv_version","alias_value":"2410.05265v2","created_at":"2026-07-05T10:05:45Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2410.05265","created_at":"2026-07-05T10:05:45Z"},{"alias_kind":"pith_short_12","alias_value":"5P27RQRK5X2O","created_at":"2026-07-05T10:05:45Z"},{"alias_kind":"pith_short_16","alias_value":"5P27RQRK5X2O3D4J","created_at":"2026-07-05T10:05:45Z"},{"alias_kind":"pith_short_8","alias_value":"5P27RQRK","created_at":"2026-07-05T10:05:45Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2024:5P27RQRK5X2O3D4JCX43JFPLFQ","target":"record","payload":{"canonical_record":{"source":{"id":"2410.05265","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-10-07T17:59:35Z","cross_cats_sorted":["cs.CL"],"title_canon_sha256":"c31cfb2b4566ce2cc2e14d25a954bc6bb14856d20c5ee49215fb18c00825923a","abstract_canon_sha256":"06338fd17b50ec208c0476c5f8bbfc132cba3138b68e71832d26f3679daa7f08"},"schema_version":"1.0"},"canonical_sha256":"ebf5f8c22aedf4ed8f8915f9b495eb2c356b9336bb7f8248324a8e05c04f2504","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:05:45.380837Z","signature_b64":"AbhqTOSMVTNMXp4H/xypyfEVJMxxtsy2KZ7NG2p/EpnEdto7oOyDX0VweBS6aXoQ75lHTW2KIfdfzGwY0RlyCw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"ebf5f8c22aedf4ed8f8915f9b495eb2c356b9336bb7f8248324a8e05c04f2504","last_reissued_at":"2026-07-05T10:05:45.380300Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:05:45.380300Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2410.05265","source_version":2,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T10:05:45Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"6cMk86OhI0UEEaoH/0GMy059/no/sHoETLSKhkTm6NTC8rg7AlsEvybu3a45iT5PsdxhSP7YxwYgkM2lboZqAg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-15T07:14:57.957786Z"},"content_sha256":"05b7d6d2795360ff6701a4a49030edc71d9324c0c78cb9f1ee943ae8521a135b","schema_version":"1.0","event_id":"sha256:05b7d6d2795360ff6701a4a49030edc71d9324c0c78cb9f1ee943ae8521a135b"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2024:5P27RQRK5X2O3D4JCX43JFPLFQ","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"PrefixQuant: Eliminating Outliers by Prefixed Tokens for Large Language Models Quantization","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.LG","authors_text":"Jiahao Wang, Mengzhao Chen, Ping Luo, Wenqi Shao, Yi Bin, Yi Liu","submitted_at":"2024-10-07T17:59:35Z","abstract_excerpt":"Existing weight-activation quantization methods for Large Language Models (LLMs) primarily address channel-wise outliers but often neglect token-wise outliers, which limits the accuracy of quantized models. In this work, we propose PrefixQuant, a novel quantization method that achieves state-of-the-art performance across various precision levels (W4A4KV4 and W4A8KV4) and granularities (dynamic and static quantization) by effectively isolating token-wise outliers. First, PrefixQuant eliminates token-wise outliers by prefixing outlier tokens in the KV cache, a process that is training-free and h"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2410.05265","kind":"arxiv","version":2},"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/2410.05265/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"},"verdict_id":null},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T10:05:45Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"Ug9IW3KimY1/Na+ZSuDeQm55jBVgYfPyLr/woTbDGdv581WN7WXA4Ff0ZZ9DTWZZv3N+YfSn4eoJg4DX7RnYDg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-15T07:14:57.959231Z"},"content_sha256":"158ad0deccafa14231df3baae04533fa6a8d214658ad15f4fe3b6eeb1d5bc7e1","schema_version":"1.0","event_id":"sha256:158ad0deccafa14231df3baae04533fa6a8d214658ad15f4fe3b6eeb1d5bc7e1"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/5P27RQRK5X2O3D4JCX43JFPLFQ/bundle.json","state_url":"https://pith.science/pith/5P27RQRK5X2O3D4JCX43JFPLFQ/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/5P27RQRK5X2O3D4JCX43JFPLFQ/bundle.json","status":"primary"}],"public_keys":[{"key_id":"pith-v1-2026-05","algorithm":"ed25519","format":"raw","public_key_b64":"stVStoiQhXFxp4s2pdzPNoqVNBMojDU/fJ2db5S3CbM=","public_key_hex":"b2d552b68890857171a78b36a5dccf368a953413288c353f7c9d9d6f94b709b3","fingerprint_sha256_b32_first128bits":"RVFV5Z2OI2J3ZUO7ERDEBCYNKS","fingerprint_sha256_hex":"8d4b5ee74e4693bcd1df2446408b0d54","rotates_at":null,"url":"https://pith.science/pith-signing-key.json","notes":"Pith uses this Ed25519 key to sign canonical record SHA-256 digests. Verify with: ed25519_verify(public_key, message=canonical_sha256_bytes, signature=base64decode(signature_b64))."}],"merge_version":"pith-open-graph-merge-v1","built_at":"2026-08-15T07:14:57Z","links":{"resolver":"https://pith.science/pith/5P27RQRK5X2O3D4JCX43JFPLFQ","bundle":"https://pith.science/pith/5P27RQRK5X2O3D4JCX43JFPLFQ/bundle.json","state":"https://pith.science/pith/5P27RQRK5X2O3D4JCX43JFPLFQ/state.json","well_known_bundle":"https://pith.science/.well-known/pith/5P27RQRK5X2O3D4JCX43JFPLFQ/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2024:5P27RQRK5X2O3D4JCX43JFPLFQ","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"06338fd17b50ec208c0476c5f8bbfc132cba3138b68e71832d26f3679daa7f08","cross_cats_sorted":["cs.CL"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-10-07T17:59:35Z","title_canon_sha256":"c31cfb2b4566ce2cc2e14d25a954bc6bb14856d20c5ee49215fb18c00825923a"},"schema_version":"1.0","source":{"id":"2410.05265","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2410.05265","created_at":"2026-07-05T10:05:45Z"},{"alias_kind":"arxiv_version","alias_value":"2410.05265v2","created_at":"2026-07-05T10:05:45Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2410.05265","created_at":"2026-07-05T10:05:45Z"},{"alias_kind":"pith_short_12","alias_value":"5P27RQRK5X2O","created_at":"2026-07-05T10:05:45Z"},{"alias_kind":"pith_short_16","alias_value":"5P27RQRK5X2O3D4J","created_at":"2026-07-05T10:05:45Z"},{"alias_kind":"pith_short_8","alias_value":"5P27RQRK","created_at":"2026-07-05T10:05:45Z"}],"graph_snapshots":[{"event_id":"sha256:158ad0deccafa14231df3baae04533fa6a8d214658ad15f4fe3b6eeb1d5bc7e1","target":"graph","created_at":"2026-07-05T10:05:45Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/2410.05265/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Existing weight-activation quantization methods for Large Language Models (LLMs) primarily address channel-wise outliers but often neglect token-wise outliers, which limits the accuracy of quantized models. In this work, we propose PrefixQuant, a novel quantization method that achieves state-of-the-art performance across various precision levels (W4A4KV4 and W4A8KV4) and granularities (dynamic and static quantization) by effectively isolating token-wise outliers. First, PrefixQuant eliminates token-wise outliers by prefixing outlier tokens in the KV cache, a process that is training-free and h","authors_text":"Jiahao Wang, Mengzhao Chen, Ping Luo, Wenqi Shao, Yi Bin, Yi Liu","cross_cats":["cs.CL"],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-10-07T17:59:35Z","title":"PrefixQuant: Eliminating Outliers by Prefixed Tokens for Large Language Models Quantization"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2410.05265","kind":"arxiv","version":2},"verdict":{"created_at":null,"id":null,"model_set":{},"one_line_summary":"","pipeline_version":null,"pith_extraction_headline":"","strongest_claim":"","weakest_assumption":""}},"verdict_id":null}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:05b7d6d2795360ff6701a4a49030edc71d9324c0c78cb9f1ee943ae8521a135b","target":"record","created_at":"2026-07-05T10:05:45Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"06338fd17b50ec208c0476c5f8bbfc132cba3138b68e71832d26f3679daa7f08","cross_cats_sorted":["cs.CL"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-10-07T17:59:35Z","title_canon_sha256":"c31cfb2b4566ce2cc2e14d25a954bc6bb14856d20c5ee49215fb18c00825923a"},"schema_version":"1.0","source":{"id":"2410.05265","kind":"arxiv","version":2}},"canonical_sha256":"ebf5f8c22aedf4ed8f8915f9b495eb2c356b9336bb7f8248324a8e05c04f2504","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"ebf5f8c22aedf4ed8f8915f9b495eb2c356b9336bb7f8248324a8e05c04f2504","first_computed_at":"2026-07-05T10:05:45.380300Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T10:05:45.380300Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"AbhqTOSMVTNMXp4H/xypyfEVJMxxtsy2KZ7NG2p/EpnEdto7oOyDX0VweBS6aXoQ75lHTW2KIfdfzGwY0RlyCw==","signature_status":"signed_v1","signed_at":"2026-07-05T10:05:45.380837Z","signed_message":"canonical_sha256_bytes"},"source_id":"2410.05265","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:05b7d6d2795360ff6701a4a49030edc71d9324c0c78cb9f1ee943ae8521a135b","sha256:158ad0deccafa14231df3baae04533fa6a8d214658ad15f4fe3b6eeb1d5bc7e1"],"state_sha256":"7c1335ecdd2027e728df0be9e4166605ff83edd089ad9f68af9056c6f92aad6c"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"EMAzEi86fuAukU0VPN1X/C5nllfap8lrKDFxYUYnKIEBCEmqHYX0W5AneEk6p00WAZNN4/hfpZslqdAMveVBAA==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-15T07:14:57.964730Z","bundle_sha256":"5c8e5549d099f7b055557b350dce5e64faede080deb398dc45d41d4d79fe443f"}}