{"id":"https://openalex.org/W7160054310","doi":"https://doi.org/10.48550/arxiv.2605.00422","title":"BWLA: Breaking the Barrier of W1AX Post-Training Quantization for LLMs","display_name":"BWLA: Breaking the Barrier of W1AX Post-Training Quantization for LLMs","publication_year":2026,"publication_date":"2026-05-01","ids":{"openalex":"https://openalex.org/W7160054310","doi":"https://doi.org/10.48550/arxiv.2605.00422"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2605.00422","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2605.00422","pdf_url":"https://arxiv.org/pdf/2605.00422","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"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":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"volume":null,"type":"preprint","indexed_in":["arxiv","datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2605.00422","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5125940974","display_name":"Zhixiong Zhao","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhao, Zhixiong","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5133307012","display_name":"Zukang Xu","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Xu, Zukang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5125938233","display_name":"Dawei Yang","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yang, Dawei","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":0,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":true,"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":0.9794999957084656,"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":0.9794999957084656,"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/T10028","display_name":"Topic Modeling","score":0.003700000001117587,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"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/T10363","display_name":"Low-power high-performance VLSI design","score":0.00139999995008111,"subfield":{"id":"https://openalex.org/subfields/2208","display_name":"Electrical and Electronic Engineering"},"field":{"id":"https://openalex.org/fields/22","display_name":"Engineering"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/post-training-quantization","display_name":"post-training quantization","score":0.9350000023841858},{"id":"https://openalex.org/keywords/large-language-models","display_name":"large language models","score":0.8299999833106995},{"id":"https://openalex.org/keywords/weight-binarization","display_name":"weight binarization","score":0.8870000243186951},{"id":"https://openalex.org/keywords/inference-acceleration","display_name":"inference acceleration","score":0.8610000014305115},{"id":"https://openalex.org/keywords/qwen3-32b","display_name":"Qwen3-32B","score":0.781000018119812}],"concepts":[{"id":"https://openalex.org/C28855332","wikidata":"https://www.wikidata.org/wiki/Q198099","display_name":"Quantization (signal processing)","level":2,"score":0.8095999956130981},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6074000000953674},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.5752999782562256},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.5629000067710876},{"id":"https://openalex.org/C100279451","wikidata":"https://www.wikidata.org/wiki/Q372193","display_name":"Perplexity","level":3,"score":0.5213000178337097},{"id":"https://openalex.org/C22789450","wikidata":"https://www.wikidata.org/wiki/Q420904","display_name":"Singular value decomposition","level":2,"score":0.4790000021457672},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.3517000079154968},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.3407000005245209},{"id":"https://openalex.org/C188198153","wikidata":"https://www.wikidata.org/wiki/Q1613840","display_name":"Limiting","level":2,"score":0.33379998803138733},{"id":"https://openalex.org/C57493831","wikidata":"https://www.wikidata.org/wiki/Q3134666","display_name":"Projection (relational algebra)","level":2,"score":0.3325999975204468},{"id":"https://openalex.org/C81081738","wikidata":"https://www.wikidata.org/wiki/Q55542","display_name":"Lossless compression","level":3,"score":0.32339999079704285},{"id":"https://openalex.org/C78548338","wikidata":"https://www.wikidata.org/wiki/Q2493","display_name":"Data compression","level":2,"score":0.31209999322891235},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.2838999927043915},{"id":"https://openalex.org/C49766605","wikidata":"https://www.wikidata.org/wiki/Q207643","display_name":"Linear map","level":2,"score":0.26100000739097595},{"id":"https://openalex.org/C180016635","wikidata":"https://www.wikidata.org/wiki/Q2712821","display_name":"Compression (physics)","level":2,"score":0.2583000063896179},{"id":"https://openalex.org/C67820243","wikidata":"https://www.wikidata.org/wiki/Q179164","display_name":"Unitary state","level":2,"score":0.2563000023365021}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2605.00422","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2605.00422","pdf_url":"https://arxiv.org/pdf/2605.00422","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"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":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2605.00422","doi":"https://doi.org/10.48550/arxiv.2605.00422","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.00422","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"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":null,"license_id":null,"version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2605.00422","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2605.00422","pdf_url":"https://arxiv.org/pdf/2605.00422","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"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":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"sustainable_development_goals":[],"study_designs":[],"awards":[],"funders":[],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7160054310.pdf","grobid_xml":"https://content.openalex.org/works/W7160054310.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W4391670689","https://openalex.org/W7212031868","https://openalex.org/W7155452190","https://openalex.org/W4417174908","https://openalex.org/W4412888573","https://openalex.org/W7164569183","https://openalex.org/W4387323141","https://openalex.org/W4403362824","https://openalex.org/W4415232763","https://openalex.org/W4409362928"],"abstract_inverted_index":{"Large":[0],"language":[1],"models":[2],"(LLMs)":[3],"have":[4],"driven":[5],"major":[6],"progress":[7],"in":[8,47],"NLP,":[9],"yet":[10],"their":[11],"substantial":[12],"memory":[13],"and":[14,31,42,63,109,155,168],"compute":[15,30],"demands":[16],"still":[17],"hinder":[18],"practical":[19],"deployment.":[20],"Binarization":[21],"can":[22],"compress":[23],"weights":[24,100],"to":[25],"1":[26],"bit,":[27],"fundamentally":[28],"lowering":[29],"bandwidth":[32],"cost.":[33],"However,":[34],"existing":[35],"methods":[36],"cannot":[37],"address":[38],"activation":[39,107],"heavy":[40],"tails":[41,108],"thus":[43],"must":[44],"keep":[45],"activations":[46,83,142],"high":[48,73],"precision,":[49],"preventing":[50],"true":[51],"end-to-end":[52],"acceleration.":[53,169],"To":[54],"overcome":[55],"this":[56],"limitation,":[57],"we":[58],"propose":[59],"BWLA":[60,133],"(Binarized":[61],"Weights":[62],"Low-bit":[64],"Activations),":[65],"the":[66],"first":[67],"post-training":[68],"quantization":[69,79],"framework":[70],"that":[71],"preserves":[72],"accuracy":[74],"while":[75,105],"achieving":[76],"1-bit":[77],"weight":[78],"together":[80],"with":[81,128],"low-bit":[82],"(e.g.,":[84],"6":[85],"bits).":[86],"The":[87,111],"Orthogonal-Kronecker":[88],"Transformation":[89],"(OKT)":[90],"learns":[91],"an":[92],"orthogonal":[93],"mapping":[94],"via":[95],"EM":[96],"minimization,":[97],"converting":[98],"unimodal":[99],"into":[101],"symmetric":[102],"bimodal":[103],"forms":[104],"suppressing":[106],"incoherence.":[110],"Proximal":[112],"SVD":[113,123],"Projection":[114],"(PSP)":[115],"then":[116],"performs":[117],"lightweight":[118],"low-rank":[119],"refinement":[120],"through":[121],"proximal":[122],"projection,":[124],"further":[125],"enhancing":[126],"quantizability":[127],"minimal":[129],"overhead.":[130],"On":[131],"Qwen3-32B,":[132],"reaches":[134],"a":[135],"Wikitext2":[136],"perplexity":[137],"of":[138],"11.92":[139],"under":[140],"6-bit":[141],"(vs.":[143],"38":[144],"from":[145],"SOTA),":[146],"improves":[147],"five":[148],"zero-shot":[149],"tasks":[150],"by":[151],"more":[152],"than":[153],"70%,":[154],"delivers":[156],"3.26":[157],"times":[158],"inference":[159],"speedup,":[160],"demonstrating":[161],"strong":[162],"potential":[163],"for":[164],"real-world":[165],"LLM":[166],"compression":[167]},"counts_by_year":[],"updated_date":"2026-10-31T12:09:11.096373","created_date":"2026-05-05T00:00:00"}