{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:TESJRXNNI5HK5F4RKLCQISRWWS","short_pith_number":"pith:TESJRXNN","schema_version":"1.0","canonical_sha256":"992498ddad474eae979152c5044a36b4b2dacd9ac694c5fff31619a813ca3b08","source":{"kind":"arxiv","id":"2506.09932","version":2},"attestation_state":"computed","paper":{"title":"HadaNorm: Diffusion Transformer Quantization through Mean-Centered Transformations","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CV","authors_text":"Boris van Breugel, Marco Federici, Markus Nagel, Paul Whatmough, Riccardo Del Chiaro","submitted_at":"2025-06-11T16:54:34Z","abstract_excerpt":"Diffusion models represent the cutting edge in image generation, but their high memory and computational demands hinder deployment on resource-constrained devices. Post-Training Quantization (PTQ) offers a promising solution by reducing the bitwidth of matrix operations. However, standard PTQ methods struggle with outliers, and achieving higher compression often requires transforming model weights and activations before quantization. In this work, we propose HadaNorm, a novel linear transformation that extends existing approaches by both normalizing channels activations and applying Hadamard t"},"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":"2506.09932","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2025-06-11T16:54:34Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"8447488e84748dbc8bd87c56487411961443f6467151f15e841ccf0212464349","abstract_canon_sha256":"5ad7e34e98c0ea4f1fdac61ef750995f2af2878a5be68e1e883f422c29e56f1e"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:34:52.051933Z","signature_b64":"hn6JOk5IUycAyGyhCMB6dQIaa1daIcbq8beoCS3atZ7+hOlLGJ6nwYcMDIUY7XjAq3suvOYyZDhi5/seAK4pAg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"992498ddad474eae979152c5044a36b4b2dacd9ac694c5fff31619a813ca3b08","last_reissued_at":"2026-07-05T11:34:52.051246Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:34:52.051246Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"HadaNorm: Diffusion Transformer Quantization through Mean-Centered Transformations","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CV","authors_text":"Boris van Breugel, Marco Federici, Markus Nagel, Paul Whatmough, Riccardo Del Chiaro","submitted_at":"2025-06-11T16:54:34Z","abstract_excerpt":"Diffusion models represent the cutting edge in image generation, but their high memory and computational demands hinder deployment on resource-constrained devices. Post-Training Quantization (PTQ) offers a promising solution by reducing the bitwidth of matrix operations. However, standard PTQ methods struggle with outliers, and achieving higher compression often requires transforming model weights and activations before quantization. In this work, we propose HadaNorm, a novel linear transformation that extends existing approaches by both normalizing channels activations and applying Hadamard t"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2506.09932","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/2506.09932/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":"2506.09932","created_at":"2026-07-05T11:34:52.051327+00:00"},{"alias_kind":"arxiv_version","alias_value":"2506.09932v2","created_at":"2026-07-05T11:34:52.051327+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2506.09932","created_at":"2026-07-05T11:34:52.051327+00:00"},{"alias_kind":"pith_short_12","alias_value":"TESJRXNNI5HK","created_at":"2026-07-05T11:34:52.051327+00:00"},{"alias_kind":"pith_short_16","alias_value":"TESJRXNNI5HK5F4R","created_at":"2026-07-05T11:34:52.051327+00:00"},{"alias_kind":"pith_short_8","alias_value":"TESJRXNN","created_at":"2026-07-05T11:34:52.051327+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/TESJRXNNI5HK5F4RKLCQISRWWS","json":"https://pith.science/pith/TESJRXNNI5HK5F4RKLCQISRWWS.json","graph_json":"https://pith.science/api/pith-number/TESJRXNNI5HK5F4RKLCQISRWWS/graph.json","events_json":"https://pith.science/api/pith-number/TESJRXNNI5HK5F4RKLCQISRWWS/events.json","paper":"https://pith.science/paper/TESJRXNN"},"agent_actions":{"view_html":"https://pith.science/pith/TESJRXNNI5HK5F4RKLCQISRWWS","download_json":"https://pith.science/pith/TESJRXNNI5HK5F4RKLCQISRWWS.json","view_paper":"https://pith.science/paper/TESJRXNN","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2506.09932&json=true","fetch_graph":"https://pith.science/api/pith-number/TESJRXNNI5HK5F4RKLCQISRWWS/graph.json","fetch_events":"https://pith.science/api/pith-number/TESJRXNNI5HK5F4RKLCQISRWWS/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/TESJRXNNI5HK5F4RKLCQISRWWS/action/timestamp_anchor","attest_storage":"https://pith.science/pith/TESJRXNNI5HK5F4RKLCQISRWWS/action/storage_attestation","attest_author":"https://pith.science/pith/TESJRXNNI5HK5F4RKLCQISRWWS/action/author_attestation","sign_citation":"https://pith.science/pith/TESJRXNNI5HK5F4RKLCQISRWWS/action/citation_signature","submit_replication":"https://pith.science/pith/TESJRXNNI5HK5F4RKLCQISRWWS/action/replication_record"}},"created_at":"2026-07-05T11:34:52.051327+00:00","updated_at":"2026-07-05T11:34:52.051327+00:00"}