{"id":"https://openalex.org/W4417261357","doi":"https://doi.org/10.48550/arxiv.2506.09932","title":"HadaNorm: Diffusion Transformer Quantization through Mean-Centered Transformations","display_name":"HadaNorm: Diffusion Transformer Quantization through Mean-Centered Transformations","publication_year":2025,"publication_date":"2025-06-11","ids":{"openalex":"https://openalex.org/W4417261357","doi":"https://doi.org/10.48550/arxiv.2506.09932"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2506.09932","is_oa":true,"landing_page_url":"http://arxiv.org/abs/2506.09932","pdf_url":"https://arxiv.org/pdf/2506.09932","source":{"id":"https://openalex.org/S4393918464","display_name":"ArXiv.org","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"type":"preprint","indexed_in":["arxiv","datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2506.09932","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5063645426","display_name":"Marco Federici","orcid":null},"institutions":[],"countries":[],"is_corresponding":true,"raw_author_name":"Federici, Marco","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5099096937","display_name":"Riccardo Del Chiaro","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Del Chiaro, Riccardo","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5018684838","display_name":"Boris van Breugel","orcid":"https://orcid.org/0009-0006-5125-0028"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"van Breugel, Boris","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5088853105","display_name":"Paul N. Whatmough","orcid":"https://orcid.org/0000-0002-1865-6492"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Whatmough, Paul","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5111645310","display_name":"Markus Nagel","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Nagel, Markus","raw_affiliation_strings":[],"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":5,"corresponding_author_ids":["https://openalex.org/A5063645426"],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":null,"last_page":null},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10036","display_name":"Advanced Neural Network Applications","score":0.3025999963283539,"subfield":{"id":"https://openalex.org/subfields/1707","display_name":"Computer Vision and Pattern Recognition"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},"topics":[{"id":"https://openalex.org/T10036","display_name":"Advanced Neural Network Applications","score":0.3025999963283539,"subfield":{"id":"https://openalex.org/subfields/1707","display_name":"Computer Vision and Pattern Recognition"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T11019","display_name":"Image Enhancement Techniques","score":0.2615000009536743,"subfield":{"id":"https://openalex.org/subfields/1707","display_name":"Computer Vision and Pattern Recognition"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T10775","display_name":"Generative Adversarial Networks and Image Synthesis","score":0.10729999840259552,"subfield":{"id":"https://openalex.org/subfields/1707","display_name":"Computer Vision and Pattern Recognition"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/quantization","display_name":"Quantization (signal processing)","score":0.6230000257492065},{"id":"https://openalex.org/keywords/outlier","display_name":"Outlier","score":0.512499988079071},{"id":"https://openalex.org/keywords/transformer","display_name":"Transformer","score":0.46939998865127563},{"id":"https://openalex.org/keywords/hadamard-transform","display_name":"Hadamard transform","score":0.38850000500679016},{"id":"https://openalex.org/keywords/linear-map","display_name":"Linear map","score":0.304500013589859},{"id":"https://openalex.org/keywords/transformation","display_name":"Transformation (genetics)","score":0.30219998955726624},{"id":"https://openalex.org/keywords/data-compression","display_name":"Data compression","score":0.30160000920295715}],"concepts":[{"id":"https://openalex.org/C28855332","wikidata":"https://www.wikidata.org/wiki/Q198099","display_name":"Quantization (signal processing)","level":2,"score":0.6230000257492065},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.6053000092506409},{"id":"https://openalex.org/C79337645","wikidata":"https://www.wikidata.org/wiki/Q779824","display_name":"Outlier","level":2,"score":0.512499988079071},{"id":"https://openalex.org/C66322947","wikidata":"https://www.wikidata.org/wiki/Q11658","display_name":"Transformer","level":3,"score":0.46939998865127563},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.4595000147819519},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.42309999465942383},{"id":"https://openalex.org/C60292330","wikidata":"https://www.wikidata.org/wiki/Q1014065","display_name":"Hadamard transform","level":2,"score":0.38850000500679016},{"id":"https://openalex.org/C49766605","wikidata":"https://www.wikidata.org/wiki/Q207643","display_name":"Linear map","level":2,"score":0.304500013589859},{"id":"https://openalex.org/C204241405","wikidata":"https://www.wikidata.org/wiki/Q461499","display_name":"Transformation (genetics)","level":3,"score":0.30219998955726624},{"id":"https://openalex.org/C78548338","wikidata":"https://www.wikidata.org/wiki/Q2493","display_name":"Data compression","level":2,"score":0.30160000920295715},{"id":"https://openalex.org/C81081738","wikidata":"https://www.wikidata.org/wiki/Q55542","display_name":"Lossless compression","level":3,"score":0.2870999872684479},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.28029999136924744},{"id":"https://openalex.org/C199833920","wikidata":"https://www.wikidata.org/wiki/Q612536","display_name":"Vector quantization","level":2,"score":0.2750999927520752},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.2732999920845032},{"id":"https://openalex.org/C106487976","wikidata":"https://www.wikidata.org/wiki/Q685816","display_name":"Matrix (chemical analysis)","level":2,"score":0.26739999651908875},{"id":"https://openalex.org/C165443888","wikidata":"https://www.wikidata.org/wiki/Q1482183","display_name":"Transformation matrix","level":3,"score":0.26460000872612},{"id":"https://openalex.org/C138236772","wikidata":"https://www.wikidata.org/wiki/Q25098575","display_name":"Edge device","level":3,"score":0.25859999656677246},{"id":"https://openalex.org/C126255220","wikidata":"https://www.wikidata.org/wiki/Q141495","display_name":"Mathematical optimization","level":1,"score":0.25440001487731934},{"id":"https://openalex.org/C47446073","wikidata":"https://www.wikidata.org/wiki/Q5165890","display_name":"Control theory (sociology)","level":3,"score":0.25360000133514404}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2506.09932","is_oa":true,"landing_page_url":"http://arxiv.org/abs/2506.09932","pdf_url":"https://arxiv.org/pdf/2506.09932","source":{"id":"https://openalex.org/S4393918464","display_name":"ArXiv.org","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2506.09932","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2506.09932","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":null,"raw_source_name":null,"raw_type":"article"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2506.09932","is_oa":true,"landing_page_url":"http://arxiv.org/abs/2506.09932","pdf_url":"https://arxiv.org/pdf/2506.09932","source":{"id":"https://openalex.org/S4393918464","display_name":"ArXiv.org","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Diffusion":[0],"models":[1],"represent":[2],"the":[3,30,96],"cutting":[4],"edge":[5],"in":[6],"image":[7],"generation,":[8],"but":[9],"their":[10],"high":[11],"memory":[12],"and":[13,42,51,74,82],"computational":[14],"demands":[15],"hinder":[16],"deployment":[17],"on":[18],"resource-constrained":[19],"devices.":[20],"Post-Training":[21],"Quantization":[22],"(PTQ)":[23],"offers":[24],"a":[25,61],"promising":[26],"solution":[27],"by":[28,69],"reducing":[29],"bitwidth":[31],"of":[32,99],"matrix":[33],"operations.":[34],"However,":[35],"standard":[36],"PTQ":[37],"methods":[38],"struggle":[39],"with":[40],"outliers,":[41],"achieving":[43],"higher":[44],"compression":[45],"often":[46],"requires":[47],"transforming":[48],"model":[49],"weights":[50],"activations":[52,73],"before":[53],"quantization.":[54,86],"In":[55],"this":[56],"work,":[57],"we":[58],"propose":[59],"HadaNorm,":[60],"novel":[62],"linear":[63],"transformation":[64],"that":[65,89],"extends":[66],"existing":[67],"approaches":[68],"both":[70],"normalizing":[71],"channels":[72],"applying":[75],"Hadamard":[76],"transforms":[77],"to":[78],"effectively":[79],"mitigate":[80],"outliers":[81],"enable":[83],"aggressive":[84],"activation":[85],"We":[87],"demonstrate":[88],"HadaNorm":[90],"consistently":[91],"reduces":[92],"quantization":[93],"error":[94],"across":[95],"various":[97],"components":[98],"transformer":[100],"blocks,":[101],"outperforming":[102],"state-of-the-art":[103],"methods.":[104]},"counts_by_year":[],"updated_date":"2026-03-07T16:01:11.037858","created_date":"2025-10-10T00:00:00"}
