{"id":"https://openalex.org/W7166887532","doi":"https://doi.org/10.18653/v1/2026.acl-long.388","title":"ARCQuant: Boosting NVFP4 Quantization with Augmented Residual Channels for LLMs","display_name":"ARCQuant: Boosting NVFP4 Quantization with Augmented Residual Channels for LLMs","publication_year":2026,"publication_date":"2026-01-01","ids":{"openalex":"https://openalex.org/W7166887532","doi":"https://doi.org/10.18653/v1/2026.acl-long.388"},"language":null,"primary_location":{"id":"doi:10.18653/v1/2026.acl-long.388","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.acl-long.388","pdf_url":"https://aclanthology.org/2026.acl-long.388.pdf","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"},"type":"conference-paper","indexed_in":["arxiv","crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://aclanthology.org/2026.acl-long.388.pdf","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5139820381","display_name":"Haoqian Meng","orcid":null},"institutions":[{"id":"https://openalex.org/I132369690","display_name":"Tianjin University of Science and Technology","ror":"https://ror.org/018rbtf37","country_code":"CN","type":"education","lineage":["https://openalex.org/I132369690"]},{"id":"https://openalex.org/I136765683","display_name":"Tianjin University of Technology","ror":"https://ror.org/00zbe0w13","country_code":"CN","type":"education","lineage":["https://openalex.org/I136765683"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Haoqian Meng","raw_affiliation_strings":["School of Computer Science and Technology , Tianjin University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"School of Computer Science and Technology , Tianjin University","institution_ids":["https://openalex.org/I132369690","https://openalex.org/I136765683"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139813173","display_name":"Yilun Luo","orcid":null},"institutions":[{"id":"https://openalex.org/I132369690","display_name":"Tianjin University of Science and Technology","ror":"https://ror.org/018rbtf37","country_code":"CN","type":"education","lineage":["https://openalex.org/I132369690"]},{"id":"https://openalex.org/I136765683","display_name":"Tianjin University of Technology","ror":"https://ror.org/00zbe0w13","country_code":"CN","type":"education","lineage":["https://openalex.org/I136765683"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Yilun Luo","raw_affiliation_strings":["School of Computer Science and Technology , Tianjin University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"School of Computer Science and Technology , Tianjin University","institution_ids":["https://openalex.org/I132369690","https://openalex.org/I136765683"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139844490","display_name":"Yafei Zhao","orcid":null},"institutions":[{"id":"https://openalex.org/I132369690","display_name":"Tianjin University of Science and Technology","ror":"https://ror.org/018rbtf37","country_code":"CN","type":"education","lineage":["https://openalex.org/I132369690"]},{"id":"https://openalex.org/I136765683","display_name":"Tianjin University of Technology","ror":"https://ror.org/00zbe0w13","country_code":"CN","type":"education","lineage":["https://openalex.org/I136765683"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Yafei Zhao","raw_affiliation_strings":["School of Computer Science and Technology , Tianjin University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"School of Computer Science and Technology , Tianjin University","institution_ids":["https://openalex.org/I132369690","https://openalex.org/I136765683"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139727837","display_name":"Wenyuan Liu","orcid":null},"institutions":[{"id":"https://openalex.org/I132369690","display_name":"Tianjin University of Science and Technology","ror":"https://ror.org/018rbtf37","country_code":"CN","type":"education","lineage":["https://openalex.org/I132369690"]},{"id":"https://openalex.org/I136765683","display_name":"Tianjin University of Technology","ror":"https://ror.org/00zbe0w13","country_code":"CN","type":"education","lineage":["https://openalex.org/I136765683"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Wenyuan Liu","raw_affiliation_strings":["School of Computer Science and Technology , Tianjin University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"School of Computer Science and Technology , Tianjin University","institution_ids":["https://openalex.org/I132369690","https://openalex.org/I136765683"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139838805","display_name":"Peng Zhang","orcid":null},"institutions":[{"id":"https://openalex.org/I132369690","display_name":"Tianjin University of Science and Technology","ror":"https://ror.org/018rbtf37","country_code":"CN","type":"education","lineage":["https://openalex.org/I132369690"]},{"id":"https://openalex.org/I136765683","display_name":"Tianjin University of Technology","ror":"https://ror.org/00zbe0w13","country_code":"CN","type":"education","lineage":["https://openalex.org/I136765683"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Peng Zhang","raw_affiliation_strings":["School of Computer Science and Technology , Tianjin University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"School of Computer Science and Technology , Tianjin University","institution_ids":["https://openalex.org/I132369690","https://openalex.org/I136765683"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5110658926","display_name":"Xindian Ma","orcid":null},"institutions":[{"id":"https://openalex.org/I132369690","display_name":"Tianjin University of Science and Technology","ror":"https://ror.org/018rbtf37","country_code":"CN","type":"education","lineage":["https://openalex.org/I132369690"]},{"id":"https://openalex.org/I136765683","display_name":"Tianjin University of Technology","ror":"https://ror.org/00zbe0w13","country_code":"CN","type":"education","lineage":["https://openalex.org/I136765683"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Xindian Ma","raw_affiliation_strings":["School of Computer Science and Technology , Tianjin University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"School of Computer Science and Technology , Tianjin University","institution_ids":["https://openalex.org/I132369690","https://openalex.org/I136765683"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":2,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":true,"cited_by_count":0,"citation_normalized_percentile":{"value":0.79534475,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"8609","last_page":"8623"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10054","display_name":"Parallel Computing and Optimization Techniques","score":0.07940000295639038,"subfield":{"id":"https://openalex.org/subfields/1708","display_name":"Hardware and Architecture"},"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/T10054","display_name":"Parallel Computing and Optimization Techniques","score":0.07940000295639038,"subfield":{"id":"https://openalex.org/subfields/1708","display_name":"Hardware and Architecture"},"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/T10036","display_name":"Advanced Neural Network Applications","score":0.04010000079870224,"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/T10901","display_name":"Advanced Data Compression Techniques","score":0.038600001484155655,"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/boosting","display_name":"Boosting (machine learning)","score":0.7383000254631042},{"id":"https://openalex.org/keywords/quantization","display_name":"Quantization (signal processing)","score":0.5504000186920166},{"id":"https://openalex.org/keywords/residual","display_name":"Residual","score":0.5242999792098999},{"id":"https://openalex.org/keywords/vector-quantization","display_name":"Vector quantization","score":0.2921000123023987},{"id":"https://openalex.org/keywords/control-theory","display_name":"Control theory (sociology)","score":0.26179999113082886}],"concepts":[{"id":"https://openalex.org/C46686674","wikidata":"https://www.wikidata.org/wiki/Q466303","display_name":"Boosting (machine learning)","level":2,"score":0.7383000254631042},{"id":"https://openalex.org/C28855332","wikidata":"https://www.wikidata.org/wiki/Q198099","display_name":"Quantization (signal processing)","level":2,"score":0.5504000186920166},{"id":"https://openalex.org/C155512373","wikidata":"https://www.wikidata.org/wiki/Q287450","display_name":"Residual","level":2,"score":0.5242999792098999},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.4887000024318695},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.38499999046325684},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.3621000051498413},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.3061000108718872},{"id":"https://openalex.org/C199833920","wikidata":"https://www.wikidata.org/wiki/Q612536","display_name":"Vector quantization","level":2,"score":0.2921000123023987},{"id":"https://openalex.org/C47446073","wikidata":"https://www.wikidata.org/wiki/Q5165890","display_name":"Control theory (sociology)","level":3,"score":0.26179999113082886},{"id":"https://openalex.org/C70153297","wikidata":"https://www.wikidata.org/wiki/Q5591907","display_name":"Gradient boosting","level":3,"score":0.2434999942779541}],"mesh":[],"locations_count":2,"locations":[{"id":"doi:10.18653/v1/2026.acl-long.388","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.acl-long.388","pdf_url":"https://aclanthology.org/2026.acl-long.388.pdf","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"},{"id":"pmh:oai:arXiv.org:2601.07475","is_oa":true,"landing_page_url":"https://arxiv.org/abs/2601.07475","pdf_url":"https://arxiv.org/pdf/2601.07475","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,"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":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"}],"best_oa_location":{"id":"doi:10.18653/v1/2026.acl-long.388","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.acl-long.388","pdf_url":"https://aclanthology.org/2026.acl-long.388.pdf","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[{"id":"https://openalex.org/G5760374524","display_name":null,"funder_award_id":"62276188","funder_id":"https://openalex.org/F4320321001","funder_display_name":"National Natural Science Foundation of China"},{"id":"https://openalex.org/G7142792453","display_name":null,"funder_award_id":"62550068","funder_id":"https://openalex.org/F4320321001","funder_display_name":"National Natural Science Foundation of China"}],"funders":[{"id":"https://openalex.org/F4320321001","display_name":"National Natural Science Foundation of China","ror":"https://ror.org/01h0zpd94"},{"id":"https://openalex.org/F4320322701","display_name":"Tianjin University","ror":"https://ror.org/012tb2g32"},{"id":"https://openalex.org/F4320337401","display_name":"Division of Emerging Frontiers","ror":"https://ror.org/01tnvpc68"}],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7166887532.pdf","grobid_xml":"https://content.openalex.org/works/W7166887532.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"The":[0],"emergence":[1],"of":[2,112,128,137],"fine-grained":[3,34],"numerical":[4],"formats":[5,140],"like":[6],"NVFP4":[7,66,86,131],"presents":[8],"new":[9],"opportunities":[10],"for":[11],"efficient":[12],"Large":[13],"Language":[14],"Model":[15],"(LLM)":[16],"inference.However,":[17],"it":[18],"is":[19,133,185],"difficult":[20],"to":[21,28,135,157,179],"adapt":[22],"existing":[23],"Post-Training":[24],"Quantization":[25],"(PTQ)":[26],"strategies":[27],"these":[29,57],"formats:":[30],"rotation-based":[31],"methods":[32,73],"compromise":[33,75],"block":[35,76],"isolation;":[36],"smoothing":[37],"techniques":[38],"struggle":[39],"with":[40,50,93,118],"significant":[41],"4-bit":[42],"quantization":[43,132],"errors;":[44],"and":[45,147,162,169],"mixed-precision":[46],"approaches":[47],"often":[48],"conflict":[49],"hardware":[51,79],"constraints":[52],"on":[53,145,166],"unified-precision":[54],"computation.To":[55],"address":[56],"challenges,":[58],"we":[59],"propose":[60],"ARC-Quant,":[61],"a":[62,83],"framework":[63],"that":[64,74,123,136,151],"boosts":[65],"performance":[67],"via":[68],"Augmented":[69],"Residual":[70],"Channels.Distinct":[71],"from":[72],"isolation":[77],"or":[78],"uniformity,":[80],"ARCQuant":[81,152],"maintains":[82],"strictly":[84],"unified":[85],"format":[87],"by":[88],"augmenting":[89],"the":[90,99,105,110,124],"activation":[91],"matrix":[92,106],"quantized":[94],"residual":[95],"channels.This":[96],"design":[97],"integrates":[98],"error":[100,126],"compensation":[101],"process":[102],"directly":[103],"into":[104],"reduction":[107],"dimension,":[108],"enabling":[109],"use":[111],"standard,":[113],"highly":[114],"optimized":[115],"GEMM":[116],"kernels":[117],"minimal":[119],"overhead.Theoretical":[120],"analysis":[121],"confirms":[122,174],"worst-case":[125],"bound":[127],"our":[129],"dual-stage":[130],"comparable":[134,156],"standard":[138],"8-bit":[139],"such":[141],"as":[142],"MXFP8.Extensive":[143],"experiments":[144],"LLaMA":[146],"Qwen":[148],"models":[149],"demonstrate":[150],"achieves":[153],"state-of-the-art":[154],"accuracy,":[155],"full-precision":[158],"baselines":[159],"in":[160],"perplexity":[161],"downstream":[163],"tasks.Furthermore,":[164],"deployment":[165],"RTX":[167,170],"5090":[168],"PRO":[171],"6000":[172],"GPUs":[173],"practical":[175],"benefits,":[176],"achieving":[177],"up":[178],"3\u00d7":[180],"speedup":[181],"over":[182],"FP16.Our":[183],"code":[184],"available":[186],"at":[187],"https://github.com/":[188],"actypedef/ARCQuant.":[189]},"counts_by_year":[],"updated_date":"2026-08-26T07:47:46.906454","created_date":"2026-07-02T00:00:00"}