{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:S4XIHWQSR3VKBZJBQ2NAXQSZYU","short_pith_number":"pith:S4XIHWQS","schema_version":"1.0","canonical_sha256":"972e83da128eeaa0e521869a0bc259c5373f03814cb7aa783827dbd3c3584de9","source":{"kind":"arxiv","id":"2504.19874","version":1},"attestation_state":"computed","paper":{"title":"TurboQuant: Online Vector Quantization with Near-optimal Distortion Rate","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.DB","cs.DS"],"primary_cat":"cs.LG","authors_text":"Amir Zandieh, Majid Daliri, Majid Hadian, Vahab Mirrokni","submitted_at":"2025-04-28T15:05:35Z","abstract_excerpt":"Vector quantization, a problem rooted in Shannon's source coding theory, aims to quantize high-dimensional Euclidean vectors while minimizing distortion in their geometric structure. We propose TurboQuant to address both mean-squared error (MSE) and inner product distortion, overcoming limitations of existing methods that fail to achieve optimal distortion rates. Our data-oblivious algorithms, suitable for online applications, achieve near-optimal distortion rates (within a small constant factor) across all bit-widths and dimensions. TurboQuant achieves this by randomly rotating input vectors,"},"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":"2504.19874","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-04-28T15:05:35Z","cross_cats_sorted":["cs.AI","cs.DB","cs.DS"],"title_canon_sha256":"588f872f21858e26ebc06fbcd148e21eaa7ef053e510632e20c7a5324f605214","abstract_canon_sha256":"065e31c31e602125c24c2343b4fe4781bf0341d013aa00bc4f4f40cd270818bc"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-20T08:03:06.476244Z","signature_b64":"Rl/tzCd7Kqpr4pB1wvoBPHnFB2gGUx0PJmPIpnXkXNB6sQMWxYLqBY5JO7dwyw6GNcpex/DBMmzhksno1NMzDQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"972e83da128eeaa0e521869a0bc259c5373f03814cb7aa783827dbd3c3584de9","last_reissued_at":"2026-05-20T08:03:06.474193Z","signature_status":"signed_v1","first_computed_at":"2026-05-20T08:03:06.474193Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"TurboQuant: Online Vector Quantization with Near-optimal Distortion Rate","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.DB","cs.DS"],"primary_cat":"cs.LG","authors_text":"Amir Zandieh, Majid Daliri, Majid Hadian, Vahab Mirrokni","submitted_at":"2025-04-28T15:05:35Z","abstract_excerpt":"Vector quantization, a problem rooted in Shannon's source coding theory, aims to quantize high-dimensional Euclidean vectors while minimizing distortion in their geometric structure. We propose TurboQuant to address both mean-squared error (MSE) and inner product distortion, overcoming limitations of existing methods that fail to achieve optimal distortion rates. Our data-oblivious algorithms, suitable for online applications, achieve near-optimal distortion rates (within a small constant factor) across all bit-widths and dimensions. TurboQuant achieves this by randomly rotating input vectors,"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2504.19874","kind":"arxiv","version":1},"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/2504.19874/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":"2504.19874","created_at":"2026-05-20T08:03:06.474306+00:00"},{"alias_kind":"arxiv_version","alias_value":"2504.19874v1","created_at":"2026-05-20T08:03:06.474306+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2504.19874","created_at":"2026-05-20T08:03:06.474306+00:00"},{"alias_kind":"pith_short_12","alias_value":"S4XIHWQSR3VK","created_at":"2026-05-20T08:03:06.474306+00:00"},{"alias_kind":"pith_short_16","alias_value":"S4XIHWQSR3VKBZJB","created_at":"2026-05-20T08:03:06.474306+00:00"},{"alias_kind":"pith_short_8","alias_value":"S4XIHWQS","created_at":"2026-05-20T08:03:06.474306+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":26,"internal_anchor_count":26,"sample":[{"citing_arxiv_id":"2605.17415","citing_title":"IVF-TQ: Calibration-Free Streaming Vector Search via a Codebook-Free Residual Layer","ref_index":31,"is_internal_anchor":true},{"citing_arxiv_id":"2605.20868","citing_title":"Runtime-Certified Bounded-Error Quantized Attention","ref_index":24,"is_internal_anchor":true},{"citing_arxiv_id":"2605.17415","citing_title":"IVF-TQ: Calibration-Free Streaming Vector Search via a Codebook-Free Residual Layer","ref_index":10,"is_internal_anchor":true},{"citing_arxiv_id":"2605.17757","citing_title":"OSCAR: Offline Spectral Covariance-Aware Rotation for 2-bit KV Cache Quantization","ref_index":36,"is_internal_anchor":true},{"citing_arxiv_id":"2605.18739","citing_title":"LongLive-2.0: An NVFP4 Parallel Infrastructure for Long Video Generation","ref_index":72,"is_internal_anchor":true},{"citing_arxiv_id":"2605.19660","citing_title":"OScaR: The Occam's Razor for Extreme KV Cache Quantization in LLMs and Beyond","ref_index":74,"is_internal_anchor":true},{"citing_arxiv_id":"2605.19972","citing_title":"Block-Sphere Vector Quantization","ref_index":4,"is_internal_anchor":true},{"citing_arxiv_id":"2605.19932","citing_title":"PEEK: Context Map as an Orientation Cache for Long-Context LLM Agents","ref_index":50,"is_internal_anchor":true},{"citing_arxiv_id":"2605.17613","citing_title":"VeriCache: Turning Lossy KV Cache into Lossless LLM Inference","ref_index":79,"is_internal_anchor":true},{"citing_arxiv_id":"2605.15507","citing_title":"PrismQuant: Rate-Distortion-Optimal Vector Quantization for Gaussian-Mixture Sources","ref_index":30,"is_internal_anchor":true},{"citing_arxiv_id":"2601.17187","citing_title":"High-Rate Quantized Matrix Multiplication I","ref_index":12,"is_internal_anchor":true},{"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":158,"is_internal_anchor":true},{"citing_arxiv_id":"2605.03562","citing_title":"HeadQ: Model-Visible Distortion and Score-Space Correction for KV-Cache Quantization","ref_index":14,"is_internal_anchor":true},{"citing_arxiv_id":"2605.08114","citing_title":"Statistical Inference and Quality Measures of KV Cache Quantisations Inspired by TurboQuant","ref_index":1,"is_internal_anchor":true},{"citing_arxiv_id":"2605.05699","citing_title":"When Quantization Is Free: An int4 KV Cache That Outruns fp16 on Apple Silicon","ref_index":1,"is_internal_anchor":true},{"citing_arxiv_id":"2605.06014","citing_title":"Quantizing With Randomized Hadamard Transforms: Efficient Heuristic Now Proven","ref_index":40,"is_internal_anchor":true},{"citing_arxiv_id":"2605.05170","citing_title":"Design Conductor 2.0: An agent builds a TurboQuant inference accelerator in 80 hours","ref_index":11,"is_internal_anchor":true},{"citing_arxiv_id":"2604.20938","citing_title":"HARBOR: Automated Harness Optimization","ref_index":43,"is_internal_anchor":true},{"citing_arxiv_id":"2604.15356","citing_title":"Sequential KV Cache Compression via Probabilistic Language Tries: Beyond the Per-Vector Shannon Limit","ref_index":13,"is_internal_anchor":true},{"citing_arxiv_id":"2604.04514","citing_title":"SuperLocalMemory V3.3: The Living Brain -- Biologically-Inspired Forgetting, Cognitive Quantization, and Multi-Channel Retrieval for Zero-LLM Agent Memory Systems","ref_index":23,"is_internal_anchor":true},{"citing_arxiv_id":"2604.05366","citing_title":"3DTurboQuant: Training-Free Near-Optimal Quantization for 3D Reconstruction Models","ref_index":5,"is_internal_anchor":true},{"citing_arxiv_id":"2605.02905","citing_title":"eOptShrinkQ: Near-Lossless KV Cache Compression Through Optimal Spectral Denoising and Quantization","ref_index":40,"is_internal_anchor":true},{"citing_arxiv_id":"2604.14442","citing_title":"Hierarchical vs. Flat Iteration in Shared-Weight Transformers","ref_index":18,"is_internal_anchor":true},{"citing_arxiv_id":"2604.16957","citing_title":"Open-TQ-Metal: Fused Compressed-Domain Attention for Long-Context LLM Inference on Apple Silicon","ref_index":12,"is_internal_anchor":true},{"citing_arxiv_id":"2604.19157","citing_title":"SAW-INT4: System-Aware 4-Bit KV-Cache Quantization for Real-World LLM Serving","ref_index":20,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/S4XIHWQSR3VKBZJBQ2NAXQSZYU","json":"https://pith.science/pith/S4XIHWQSR3VKBZJBQ2NAXQSZYU.json","graph_json":"https://pith.science/api/pith-number/S4XIHWQSR3VKBZJBQ2NAXQSZYU/graph.json","events_json":"https://pith.science/api/pith-number/S4XIHWQSR3VKBZJBQ2NAXQSZYU/events.json","paper":"https://pith.science/paper/S4XIHWQS"},"agent_actions":{"view_html":"https://pith.science/pith/S4XIHWQSR3VKBZJBQ2NAXQSZYU","download_json":"https://pith.science/pith/S4XIHWQSR3VKBZJBQ2NAXQSZYU.json","view_paper":"https://pith.science/paper/S4XIHWQS","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2504.19874&json=true","fetch_graph":"https://pith.science/api/pith-number/S4XIHWQSR3VKBZJBQ2NAXQSZYU/graph.json","fetch_events":"https://pith.science/api/pith-number/S4XIHWQSR3VKBZJBQ2NAXQSZYU/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/S4XIHWQSR3VKBZJBQ2NAXQSZYU/action/timestamp_anchor","attest_storage":"https://pith.science/pith/S4XIHWQSR3VKBZJBQ2NAXQSZYU/action/storage_attestation","attest_author":"https://pith.science/pith/S4XIHWQSR3VKBZJBQ2NAXQSZYU/action/author_attestation","sign_citation":"https://pith.science/pith/S4XIHWQSR3VKBZJBQ2NAXQSZYU/action/citation_signature","submit_replication":"https://pith.science/pith/S4XIHWQSR3VKBZJBQ2NAXQSZYU/action/replication_record"}},"created_at":"2026-05-20T08:03:06.474306+00:00","updated_at":"2026-05-20T08:03:06.474306+00:00"}