{"id":"https://openalex.org/W4412887920","doi":"https://doi.org/10.18653/v1/2025.findings-acl.943","title":"FPE2M2: Approaching Lossless and Efficient Quantization with Native Floating Point","display_name":"FPE2M2: Approaching Lossless and Efficient Quantization with Native Floating Point","publication_year":2025,"publication_date":"2025-01-01","ids":{"openalex":"https://openalex.org/W4412887920","doi":"https://doi.org/10.18653/v1/2025.findings-acl.943"},"language":"en","primary_location":{"id":"doi:10.18653/v1/2025.findings-acl.943","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-acl.943","pdf_url":"https://aclanthology.org/2025.findings-acl.943.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":"Findings of the Association for Computational Linguistics: ACL 2025","raw_type":"proceedings-article"},"type":"conference-paper","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://aclanthology.org/2025.findings-acl.943.pdf","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5101459890","display_name":"Ke Yi","orcid":"https://orcid.org/0000-0002-1331-2703"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ke Yi","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5046286884","display_name":"Jianwei Zhang","orcid":"https://orcid.org/0000-0003-4533-6264"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jianwei Zhang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5101699041","display_name":"Zhiying Xu","orcid":"https://orcid.org/0000-0003-0233-9713"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhiying Xu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5109640507","display_name":"Xinlong Yang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Xinlong Yang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100460020","display_name":"Zhou Yang","orcid":"https://orcid.org/0000-0001-5203-8199"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yang Zhou","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5101927978","display_name":"Minmin Sun","orcid":"https://orcid.org/0000-0003-1923-5208"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Minmin Sun","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Zengke Liu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zengke Liu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100378796","display_name":"Tong Zhang","orcid":"https://orcid.org/0000-0002-6434-0712"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Tong Zhang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100612233","display_name":"Junyang Lin","orcid":"https://orcid.org/0000-0001-9931-383X"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Junyang Lin","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5057864403","display_name":"Jingren Zhou","orcid":"https://orcid.org/0000-0002-4220-2634"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jingren Zhou","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":"18350","last_page":"18361"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T12611","display_name":"Neural Networks and Reservoir Computing","score":0.7192999720573425,"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"}},"topics":[{"id":"https://openalex.org/T12611","display_name":"Neural Networks and Reservoir Computing","score":0.7192999720573425,"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/T11034","display_name":"Digital Filter Design and Implementation","score":0.6808000206947327,"subfield":{"id":"https://openalex.org/subfields/1711","display_name":"Signal Processing"},"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/T10299","display_name":"Photonic and Optical Devices","score":0.675000011920929,"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/lossless-compression","display_name":"Lossless compression","score":0.7736911177635193},{"id":"https://openalex.org/keywords/quantization","display_name":"Quantization (signal processing)","score":0.7034293413162231},{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.6291148662567139},{"id":"https://openalex.org/keywords/algorithm","display_name":"Algorithm","score":0.2152782380580902},{"id":"https://openalex.org/keywords/data-compression","display_name":"Data compression","score":0.21521621942520142}],"concepts":[{"id":"https://openalex.org/C81081738","wikidata":"https://www.wikidata.org/wiki/Q55542","display_name":"Lossless compression","level":3,"score":0.7736911177635193},{"id":"https://openalex.org/C28855332","wikidata":"https://www.wikidata.org/wiki/Q198099","display_name":"Quantization (signal processing)","level":2,"score":0.7034293413162231},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6291148662567139},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.2152782380580902},{"id":"https://openalex.org/C78548338","wikidata":"https://www.wikidata.org/wiki/Q2493","display_name":"Data compression","level":2,"score":0.21521621942520142}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.18653/v1/2025.findings-acl.943","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-acl.943","pdf_url":"https://aclanthology.org/2025.findings-acl.943.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":"Findings of the Association for Computational Linguistics: ACL 2025","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.18653/v1/2025.findings-acl.943","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-acl.943","pdf_url":"https://aclanthology.org/2025.findings-acl.943.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":"Findings of the Association for Computational Linguistics: ACL 2025","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4412887920.pdf","grobid_xml":"https://content.openalex.org/works/W4412887920.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W4391375266","https://openalex.org/W2899084033","https://openalex.org/W2748952813","https://openalex.org/W3106969033","https://openalex.org/W2186939576","https://openalex.org/W2357988910","https://openalex.org/W2948148442","https://openalex.org/W2377158164","https://openalex.org/W2187600494","https://openalex.org/W2114030128"],"abstract_inverted_index":{"Auto-regressive":[0],"decoding":[1,7],"is":[2,10,24],"a":[3,25,94,183],"memory-bound":[4,31],"job,":[5],"meaning":[6],"inference":[8,160,185],"performance":[9,121],"limited":[11],"by":[12],"the":[13,17,21,30,47,71,75,84,90,103,119,136,144,152,163],"bandwidth":[14],"rather":[15],"than":[16],"computational":[18],"capabilities":[19],"of":[20,37,51,97,125,131],"GPU.Weight-only":[22],"quantization":[23,44,77,126,159],"promising":[26],"method":[27],"to":[28,64,74,92,107],"address":[29],"limitations.Previous":[32],"studies":[33],"have":[34,40,54],"followed":[35],"one":[36],"two":[38],"approaches.Some":[39],"exclusively":[41],"studied":[42],"integer":[43],"while":[45,187],"ignoring":[46],"Gaussian":[48],"distribution":[49],"nature":[50],"LLMs'":[52],"weights.Others":[53],"proposed":[55],"nonuniform":[56],"quantization,":[57],"but":[58],"incurred":[59],"additional":[60],"memory":[61],"overhead":[62],"due":[63],"lookup":[65],"tables.In":[66],"this":[67],"work,":[68],"we":[69,100,128],"extend":[70],"floating-point":[72],"standard":[73],"ExMy":[76,106,133],"schema,":[78],"which":[79,115,181],"allocates":[80],"x":[81],"bits":[82,88],"for":[83,89],"exponent":[85],"and":[86,161,168,178],"y":[87],"mantissa":[91],"represent":[93],"number.In":[95],"terms":[96,124],"runtime":[98],"efficiency,":[99],"demonstrate":[101],"that":[102,130,155],"conversion":[104],"from":[105],"FP16":[108],"can":[109,116],"be":[110],"realized":[111],"through":[112],"register-level":[113],"operations,":[114],"get":[117],"almost":[118],"same":[120],"as":[122,175],"INT5.In":[123],"loss,":[127],"analyze":[129],"different":[132],"settings,":[134],"where":[135],"E2M2":[137],"schema":[138],"achieves":[139,182],"an":[140],"optimal":[141],"balance,":[142],"offering":[143],"highest":[145],"efficiency":[146],"with":[147],"lossless":[148,157,190],"accuracy.We":[149],"further":[150],"propose":[151],"FPE2M2":[153,164],"framework":[154,165],"supports":[156],"weight-only":[158],"validate":[162],"on":[166],"Qwen":[167],"LLaMA":[169],"Models":[170],"across":[171],"various":[172],"modalities,":[173],"such":[174],"text,":[176],"image,":[177],"audio":[179],"tasks,":[180],"faster":[184],"speed":[186],"maintaining":[188],"nearly":[189],"accuracy.":[191],"1":[192]},"counts_by_year":[],"updated_date":"2026-07-14T23:27:15.235271","created_date":"2025-10-10T00:00:00"}