{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:KAFFW3OSV2CDZULTK42RSIEY26","short_pith_number":"pith:KAFFW3OS","schema_version":"1.0","canonical_sha256":"500a5b6dd2ae843cd1735735192098d7aaa13e1a631f45d46210cb0cebc8e0e3","source":{"kind":"arxiv","id":"2606.23406","version":1},"attestation_state":"computed","paper":{"title":"HyperQuant: A Rate-Distortion-Optimal Quantization Pipeline for Large Language and Diffusion Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Hadar Sackstein, Tomer Solberg, Yuval Domb","submitted_at":"2026-06-22T14:30:47Z","abstract_excerpt":"We present HyperQuant (Hadamard, optimallY Packing, Entropy Rice-coding), a unified post-training quantization pipeline for the weights and the KV cache of large language and diffusion transformers. Across a suite of self-contained experiments (Table 1), HyperQuant outperforms the recent HIGGS scheme at every operating point from 3 to 5 bits per scalar (bps) on weights, and beats both TurboQuant and OCTOPUS on KV quantization down to 1.7 bps. Beyond the LLM setting, HyperQuant quantizes the 19B-parameter LTX-2 DiT video model with no observable per-frame artifacts. End-to-end on an H100 at 4 b"},"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":"2606.23406","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-06-22T14:30:47Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"78c54351bc599e9367f7b6fef038a0f54604ea9b77f75f1120325c433e707d89","abstract_canon_sha256":"591d4064663aa296e76c77be20d0753541ee46b4692e3496a88b19d5e2d87e98"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-23T03:14:19.227919Z","signature_b64":"PZTDDpNeY2U5rOOEY95Th1Tiz74rR/QxC0db3NSDLpWTUtdTV4UrTnTpuXJRYDn/5631cKUUIvrv7BrT3sFHCw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"500a5b6dd2ae843cd1735735192098d7aaa13e1a631f45d46210cb0cebc8e0e3","last_reissued_at":"2026-06-23T03:14:19.227451Z","signature_status":"signed_v1","first_computed_at":"2026-06-23T03:14:19.227451Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"HyperQuant: A Rate-Distortion-Optimal Quantization Pipeline for Large Language and Diffusion Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Hadar Sackstein, Tomer Solberg, Yuval Domb","submitted_at":"2026-06-22T14:30:47Z","abstract_excerpt":"We present HyperQuant (Hadamard, optimallY Packing, Entropy Rice-coding), a unified post-training quantization pipeline for the weights and the KV cache of large language and diffusion transformers. Across a suite of self-contained experiments (Table 1), HyperQuant outperforms the recent HIGGS scheme at every operating point from 3 to 5 bits per scalar (bps) on weights, and beats both TurboQuant and OCTOPUS on KV quantization down to 1.7 bps. Beyond the LLM setting, HyperQuant quantizes the 19B-parameter LTX-2 DiT video model with no observable per-frame artifacts. End-to-end on an H100 at 4 b"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.23406","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/2606.23406/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":"2606.23406","created_at":"2026-06-23T03:14:19.227512+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.23406v1","created_at":"2026-06-23T03:14:19.227512+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.23406","created_at":"2026-06-23T03:14:19.227512+00:00"},{"alias_kind":"pith_short_12","alias_value":"KAFFW3OSV2CD","created_at":"2026-06-23T03:14:19.227512+00:00"},{"alias_kind":"pith_short_16","alias_value":"KAFFW3OSV2CDZULT","created_at":"2026-06-23T03:14:19.227512+00:00"},{"alias_kind":"pith_short_8","alias_value":"KAFFW3OS","created_at":"2026-06-23T03:14:19.227512+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/KAFFW3OSV2CDZULTK42RSIEY26","json":"https://pith.science/pith/KAFFW3OSV2CDZULTK42RSIEY26.json","graph_json":"https://pith.science/api/pith-number/KAFFW3OSV2CDZULTK42RSIEY26/graph.json","events_json":"https://pith.science/api/pith-number/KAFFW3OSV2CDZULTK42RSIEY26/events.json","paper":"https://pith.science/paper/KAFFW3OS"},"agent_actions":{"view_html":"https://pith.science/pith/KAFFW3OSV2CDZULTK42RSIEY26","download_json":"https://pith.science/pith/KAFFW3OSV2CDZULTK42RSIEY26.json","view_paper":"https://pith.science/paper/KAFFW3OS","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.23406&json=true","fetch_graph":"https://pith.science/api/pith-number/KAFFW3OSV2CDZULTK42RSIEY26/graph.json","fetch_events":"https://pith.science/api/pith-number/KAFFW3OSV2CDZULTK42RSIEY26/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/KAFFW3OSV2CDZULTK42RSIEY26/action/timestamp_anchor","attest_storage":"https://pith.science/pith/KAFFW3OSV2CDZULTK42RSIEY26/action/storage_attestation","attest_author":"https://pith.science/pith/KAFFW3OSV2CDZULTK42RSIEY26/action/author_attestation","sign_citation":"https://pith.science/pith/KAFFW3OSV2CDZULTK42RSIEY26/action/citation_signature","submit_replication":"https://pith.science/pith/KAFFW3OSV2CDZULTK42RSIEY26/action/replication_record"}},"created_at":"2026-06-23T03:14:19.227512+00:00","updated_at":"2026-06-23T03:14:19.227512+00:00"}