{"id":"https://openalex.org/W7204156019","doi":"https://doi.org/10.48550/arxiv.2608.21952","title":"SSDi8: Accurate and Efficient 8-bit Quantization for State Space Duality","display_name":"SSDi8: Accurate and Efficient 8-bit Quantization for State Space Duality","publication_year":2026,"publication_date":"2026-08-22","ids":{"openalex":"https://openalex.org/W7204156019","doi":"https://doi.org/10.48550/arxiv.2608.21952"},"language":null,"primary_location":{"id":"pmh:oai:arXiv.org:2608.21952","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2608.21952","pdf_url":"https://arxiv.org/pdf/2608.21952","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/2608.21952","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5148130132","display_name":"Hyunwoo Kim","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Kim, Hyunwoo","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5148108727","display_name":"Byoungchan Ko","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ko, Byoungchan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5076023548","display_name":"\uac15\ubbfc\uc11d","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Kang, Minseok","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100630438","display_name":"Minwoo Kim","orcid":"https://orcid.org/0000-0001-7547-2596","observed_orcids":["https://orcid.org/0000-0001-7547-2596"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Kim, Minwoo","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5148142090","display_name":"Dongjin Lee","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Lee, Dongjin","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Lee, Jaehoon","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Lee, Jaehoon","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5086877012","display_name":"Sungroh Yoon","orcid":"https://orcid.org/0000-0002-2367-197X","observed_orcids":["https://orcid.org/0000-0002-2367-197X"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yoon, Sungroh","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5010990289","display_name":"Dahuin Jung","orcid":"https://orcid.org/0000-0002-1344-1054","observed_orcids":["https://orcid.org/0000-0002-1344-1054"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jung, Dahuin","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":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":0.9916999936103821,"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":0.9916999936103821,"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/T10363","display_name":"Low-power high-performance VLSI design","score":0.0008999999845400453,"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"}},{"id":"https://openalex.org/T11697","display_name":"Numerical Methods and Algorithms","score":0.0006000000284984708,"subfield":{"id":"https://openalex.org/subfields/1703","display_name":"Computational Theory and Mathematics"},"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/8-bit-quantization","display_name":"8-bit quantization","score":0.7910000085830688},{"id":"https://openalex.org/keywords/mamba-2","display_name":"Mamba-2","score":0.8299999833106995},{"id":"https://openalex.org/keywords/post-training-quantization","display_name":"post-training quantization","score":0.7429999709129333},{"id":"https://openalex.org/keywords/int8","display_name":"INT8","score":0.47600001096725464},{"id":"https://openalex.org/keywords/error-correction","display_name":"error correction","score":0.5329999923706055},{"id":"https://openalex.org/keywords/long-range-dependency-modeling","display_name":"long-range dependency modeling","score":0.6340000033378601},{"id":"https://openalex.org/keywords/sequence-modeling","display_name":"sequence modeling","score":0.5680000185966492},{"id":"https://openalex.org/keywords/resource-constrained-deployment","display_name":"resource-constrained deployment","score":0.6179999709129333}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6402000188827515},{"id":"https://openalex.org/C28855332","wikidata":"https://www.wikidata.org/wiki/Q198099","display_name":"Quantization (signal processing)","level":2,"score":0.6144999861717224},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.5623000264167786},{"id":"https://openalex.org/C63479239","wikidata":"https://www.wikidata.org/wiki/Q7353546","display_name":"Robustness (evolution)","level":3,"score":0.5580999851226807},{"id":"https://openalex.org/C68339613","wikidata":"https://www.wikidata.org/wiki/Q1549489","display_name":"Speedup","level":2,"score":0.5163000226020813},{"id":"https://openalex.org/C206588197","wikidata":"https://www.wikidata.org/wiki/Q846574","display_name":"Reuse","level":2,"score":0.5120000243186951},{"id":"https://openalex.org/C2778023678","wikidata":"https://www.wikidata.org/wiki/Q554403","display_name":"Duality (order theory)","level":2,"score":0.4661000072956085},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.447299987077713},{"id":"https://openalex.org/C79337645","wikidata":"https://www.wikidata.org/wiki/Q779824","display_name":"Outlier","level":2,"score":0.4253000020980835},{"id":"https://openalex.org/C42355184","wikidata":"https://www.wikidata.org/wiki/Q1361088","display_name":"Matrix decomposition","level":3,"score":0.3630000054836273},{"id":"https://openalex.org/C62799726","wikidata":"https://www.wikidata.org/wiki/Q190056","display_name":"Hilbert space","level":2,"score":0.3587999939918518},{"id":"https://openalex.org/C17349429","wikidata":"https://www.wikidata.org/wiki/Q1049914","display_name":"Matrix multiplication","level":3,"score":0.3391000032424927},{"id":"https://openalex.org/C103088060","wikidata":"https://www.wikidata.org/wiki/Q1062839","display_name":"Error detection and correction","level":2,"score":0.3343000113964081},{"id":"https://openalex.org/C72434380","wikidata":"https://www.wikidata.org/wiki/Q230930","display_name":"State space","level":2,"score":0.32280001044273376},{"id":"https://openalex.org/C48044578","wikidata":"https://www.wikidata.org/wiki/Q727490","display_name":"Scalability","level":2,"score":0.30970001220703125},{"id":"https://openalex.org/C48103436","wikidata":"https://www.wikidata.org/wiki/Q599031","display_name":"State (computer science)","level":2,"score":0.30959999561309814},{"id":"https://openalex.org/C149810388","wikidata":"https://www.wikidata.org/wiki/Q5374873","display_name":"Emulation","level":2,"score":0.3061999976634979},{"id":"https://openalex.org/C113775141","wikidata":"https://www.wikidata.org/wiki/Q428691","display_name":"Computer engineering","level":1,"score":0.29600000381469727},{"id":"https://openalex.org/C78548338","wikidata":"https://www.wikidata.org/wiki/Q2493","display_name":"Data compression","level":2,"score":0.2903999984264374},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.2732999920845032},{"id":"https://openalex.org/C104267543","wikidata":"https://www.wikidata.org/wiki/Q208163","display_name":"Signal processing","level":3,"score":0.2612999975681305},{"id":"https://openalex.org/C2778572836","wikidata":"https://www.wikidata.org/wiki/Q380933","display_name":"Space (punctuation)","level":2,"score":0.26080000400543213}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2608.21952","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2608.21952","pdf_url":"https://arxiv.org/pdf/2608.21952","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.2608.21952","doi":"https://doi.org/10.48550/arxiv.2608.21952","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2608.21952","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":null,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2608.21952","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2608.21952","pdf_url":"https://arxiv.org/pdf/2608.21952","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":[],"sustainable_development_goals_aurora":[],"study_designs":[],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Recent":[0],"advances":[1],"in":[2,164,174],"sequence":[3],"modeling":[4,17],"have":[5],"highlighted":[6],"Mamba":[7],"as":[8],"a":[9,20,84,90],"state":[10],"space":[11],"architecture":[12],"offering":[13],"efficient":[14,61],"long-range":[15],"dependency":[16],"and":[18,39,45,54,137,166],"providing":[19],"viable":[21],"alternative":[22],"to":[23,42,65,82,156,161],"Transformers.":[24],"Building":[25],"upon":[26],"this,":[27],"Mamba-2":[28],"introduces":[29,89],"the":[30,58,73,119,125,181],"Structured":[31],"State":[32],"Space":[33],"Duality":[34],"(SSD),":[35],"which":[36],"integrates":[37],"recurrent":[38],"attention":[40],"modes":[41],"achieve":[43],"efficiency":[44],"scalability.":[46],"However,":[47],"this":[48,68],"architectural":[49],"expansion":[50],"substantially":[51],"increases":[52],"memory":[53],"latency":[55],"overhead,":[56],"underscoring":[57],"need":[59],"for":[60,80],"compression":[62],"strategies":[63],"tailored":[64],"SSD.":[66],"In":[67],"work,":[69],"we":[70],"present":[71],"SSDi8,":[72],"first":[74],"post-training":[75],"quantization":[76],"framework":[77],"specifically":[78],"designed":[79],"SSD":[81],"maintain":[83],"persistent":[85],"INT8":[86],"path.":[87],"SSDi8":[88,107,122,152],"reformulation":[91],"that":[92,151],"decouples":[93],"element-wise":[94],"multiplications":[95],"from":[96],"matrix":[97],"multiplications,":[98],"enabling":[99],"reuse":[100],"of":[101,129],"quantized":[102],"activations":[103,111],"across":[104,135],"modules.":[105],"Moreover,":[106],"adaptively":[108],"quantizes":[109],"channel-varying":[110],"at":[112],"cost-effective":[113],"points,":[114],"further":[115,170],"reducing":[116],"latency.":[117],"On":[118],"accuracy":[120,154],"side,":[121],"explicitly":[123],"leverages":[124],"intrinsic":[126],"dimensional":[127],"decomposition":[128],"SSD,":[130],"exploiting":[131],"distinct":[132],"outlier":[133],"distributions":[134],"axes,":[136],"incorporates":[138],"an":[139],"error":[140,146],"correction":[141],"term":[142],"based":[143],"on":[144,180],"per-channel":[145],"statistics.":[147],"Comprehensive":[148],"experiments":[149],"demonstrate":[150],"achieves":[153],"comparable":[155],"FP16":[157],"while":[158],"delivering":[159],"up":[160],"1.4x":[162],"speedup":[163],"W4A8":[165],"W8A8":[167],"settings.":[168],"We":[169],"validate":[171],"its":[172],"robustness":[173],"resource-constrained":[175],"environments":[176],"by":[177],"deploying":[178],"it":[179],"Orin":[182],"NX":[183],"device.":[184]},"counts_by_year":[],"updated_date":"2026-10-09T09:14:12.587421","created_date":"2026-08-26T00:00:00"}