{"id":"https://openalex.org/W4415270764","doi":"https://doi.org/10.1145/3767295.3803580","title":"Laminar: A Scalable Asynchronous RL Post-Training Framework","display_name":"Laminar: A Scalable Asynchronous RL Post-Training Framework","publication_year":2026,"publication_date":"2026-04-24","ids":{"openalex":"https://openalex.org/W4415270764","doi":"https://doi.org/10.1145/3767295.3803580"},"language":"en","primary_location":{"id":"doi:10.1145/3767295.3803580","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3767295.3803580","pdf_url":null,"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 21st European Conference on Computer Systems","raw_type":"proceedings-article"},"type":"article","indexed_in":["arxiv","crossref","datacite"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://doi.org/10.1145/3767295.3803580","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5020661868","display_name":"Guangming Sheng","orcid":"https://orcid.org/0000-0003-3395-3994"},"institutions":[{"id":"https://openalex.org/I200769079","display_name":"Hong Kong University of Science and Technology","ror":"https://ror.org/00q4vv597","country_code":"HK","type":"education","lineage":["https://openalex.org/I200769079"]},{"id":"https://openalex.org/I889458895","display_name":"University of Hong Kong","ror":"https://ror.org/02zhqgq86","country_code":"HK","type":"education","lineage":["https://openalex.org/I889458895"]}],"countries":["HK"],"is_corresponding":false,"raw_author_name":"Guangming Sheng","raw_affiliation_strings":["Computer Science, The University of Hong Kong, Hong Kong, Hong Kong"],"raw_orcid":"https://orcid.org/0000-0003-3395-3994","affiliations":[{"raw_affiliation_string":"Computer Science, The University of Hong Kong, Hong Kong, Hong Kong","institution_ids":["https://openalex.org/I200769079","https://openalex.org/I889458895"]}]},{"author_position":"middle","author":{"id":null,"display_name":"Yuxuan Tong","orcid":"https://orcid.org/0009-0002-5758-5954"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yuxuan Tong","raw_affiliation_strings":["ByteDance, Beijing, China"],"raw_orcid":"https://orcid.org/0009-0002-5758-5954","affiliations":[{"raw_affiliation_string":"ByteDance, Beijing, China","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5093967737","display_name":"Borui Wan","orcid":"https://orcid.org/0009-0008-5902-1611"},"institutions":[{"id":"https://openalex.org/I200769079","display_name":"Hong Kong University of Science and Technology","ror":"https://ror.org/00q4vv597","country_code":"HK","type":"education","lineage":["https://openalex.org/I200769079"]},{"id":"https://openalex.org/I889458895","display_name":"University of Hong Kong","ror":"https://ror.org/02zhqgq86","country_code":"HK","type":"education","lineage":["https://openalex.org/I889458895"]}],"countries":["HK"],"is_corresponding":false,"raw_author_name":"Borui Wan","raw_affiliation_strings":["Computer Science, The University of Hong Kong, Hong Kong, Hong Kong"],"raw_orcid":"https://orcid.org/0009-0008-5902-1611","affiliations":[{"raw_affiliation_string":"Computer Science, The University of Hong Kong, Hong Kong, Hong Kong","institution_ids":["https://openalex.org/I200769079","https://openalex.org/I889458895"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5015520760","display_name":"Wang Zhang","orcid":"https://orcid.org/0009-0008-6043-6574"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wang Zhang","raw_affiliation_strings":["ByteDance Inc., San Jose, USA"],"raw_orcid":"https://orcid.org/0009-0001-2436-8761","affiliations":[{"raw_affiliation_string":"ByteDance Inc., San Jose, USA","institution_ids":[]}]},{"author_position":"middle","author":{"id":null,"display_name":"Chaobo Jia","orcid":"https://orcid.org/0009-0005-3309-9179"},"institutions":[{"id":"https://openalex.org/I177725633","display_name":"Chinese University of Hong Kong","ror":"https://ror.org/00t33hh48","country_code":"HK","type":"education","lineage":["https://openalex.org/I177725633"]}],"countries":["HK"],"is_corresponding":false,"raw_author_name":"Chaobo Jia","raw_affiliation_strings":["Computer Science and Engineering, The Chinese University of Hong Kong, Hong Kong, Hong Kong"],"raw_orcid":"https://orcid.org/0009-0005-3309-9179","affiliations":[{"raw_affiliation_string":"Computer Science and Engineering, The Chinese University of Hong Kong, Hong Kong, Hong Kong","institution_ids":["https://openalex.org/I177725633"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5063756360","display_name":"Xibin Wu","orcid":null},"institutions":[{"id":"https://openalex.org/I4210123185","display_name":"Zhejiang Lab","ror":"https://ror.org/02m2h7991","country_code":"CN","type":"facility","lineage":["https://openalex.org/I4210123185"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Xibin Wu","raw_affiliation_strings":["Bytedance, Hangzhou, China"],"raw_orcid":"https://orcid.org/0009-0007-9702-5288","affiliations":[{"raw_affiliation_string":"Bytedance, Hangzhou, China","institution_ids":["https://openalex.org/I4210123185"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5044348233","display_name":"Yuqi Wu","orcid":"https://orcid.org/0000-0001-7949-7838"},"institutions":[{"id":"https://openalex.org/I183067930","display_name":"Shanghai Jiao Tong University","ror":"https://ror.org/0220qvk04","country_code":"CN","type":"education","lineage":["https://openalex.org/I183067930"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Yuqi Wu","raw_affiliation_strings":["Shanghai Jiaotong University, Shanghai, China"],"raw_orcid":"https://orcid.org/0009-0004-6144-7399","affiliations":[{"raw_affiliation_string":"Shanghai Jiaotong University, Shanghai, China","institution_ids":["https://openalex.org/I183067930"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100745285","display_name":"Xiang Li","orcid":"https://orcid.org/0009-0009-0556-5290"},"institutions":[{"id":"https://openalex.org/I150744194","display_name":"Waseda University","ror":"https://ror.org/00ntfnx83","country_code":"JP","type":"education","lineage":["https://openalex.org/I150744194"]}],"countries":["JP"],"is_corresponding":false,"raw_author_name":"Xiang Li","raw_affiliation_strings":["Bytedance, Tokyo, Japan"],"raw_orcid":"https://orcid.org/0009-0002-8751-4655","affiliations":[{"raw_affiliation_string":"Bytedance, Tokyo, Japan","institution_ids":["https://openalex.org/I150744194"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100458134","display_name":"Chi Zhang","orcid":"https://orcid.org/0000-0001-7374-1940"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Chi Zhang","raw_affiliation_strings":["ByteDance, Shanghai, China"],"raw_orcid":"https://orcid.org/0000-0001-7374-1940","affiliations":[{"raw_affiliation_string":"ByteDance, Shanghai, China","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5013856812","display_name":"Yanghua Peng","orcid":"https://orcid.org/0000-0003-3989-4358"},"institutions":[{"id":"https://openalex.org/I58610484","display_name":"Seattle University","ror":"https://ror.org/02jqc0m91","country_code":"US","type":"education","lineage":["https://openalex.org/I58610484"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Yanghua Peng","raw_affiliation_strings":["ByteDance Inc., Seattle, USA"],"raw_orcid":"https://orcid.org/0000-0003-3989-4358","affiliations":[{"raw_affiliation_string":"ByteDance Inc., Seattle, USA","institution_ids":["https://openalex.org/I58610484"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5084047386","display_name":"Haibin Lin","orcid":"https://orcid.org/0000-0003-4879-5335"},"institutions":[{"id":"https://openalex.org/I58610484","display_name":"Seattle University","ror":"https://ror.org/02jqc0m91","country_code":"US","type":"education","lineage":["https://openalex.org/I58610484"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Haibin Lin","raw_affiliation_strings":["ByteDance Inc., Seattle, USA"],"raw_orcid":"https://orcid.org/0000-0003-4879-5335","affiliations":[{"raw_affiliation_string":"ByteDance Inc., Seattle, USA","institution_ids":["https://openalex.org/I58610484"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5113343221","display_name":"Xin Liu","orcid":"https://orcid.org/0009-0000-8346-3323"},"institutions":[{"id":"https://openalex.org/I58610484","display_name":"Seattle University","ror":"https://ror.org/02jqc0m91","country_code":"US","type":"education","lineage":["https://openalex.org/I58610484"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Xin Liu","raw_affiliation_strings":["ByteDance Inc., Seattle, USA"],"raw_orcid":"https://orcid.org/0009-0000-8346-3323","affiliations":[{"raw_affiliation_string":"ByteDance Inc., Seattle, USA","institution_ids":["https://openalex.org/I58610484"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5101232348","display_name":"Chuan Wu","orcid":null},"institutions":[{"id":"https://openalex.org/I200769079","display_name":"Hong Kong University of Science and Technology","ror":"https://ror.org/00q4vv597","country_code":"HK","type":"education","lineage":["https://openalex.org/I200769079"]},{"id":"https://openalex.org/I889458895","display_name":"University of Hong Kong","ror":"https://ror.org/02zhqgq86","country_code":"HK","type":"education","lineage":["https://openalex.org/I889458895"]}],"countries":["HK"],"is_corresponding":false,"raw_author_name":"Chuan Wu","raw_affiliation_strings":["Computer Science, The University of Hong Kong, Hong Kong, Hong Kong"],"raw_orcid":"https://orcid.org/0000-0002-3144-4398","affiliations":[{"raw_affiliation_string":"Computer Science, The University of Hong Kong, Hong Kong, Hong Kong","institution_ids":["https://openalex.org/I200769079","https://openalex.org/I889458895"]}]}],"institutions":[],"countries_distinct_count":4,"institutions_distinct_count":7,"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":0.00687285,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"400","last_page":"422"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10054","display_name":"Parallel Computing and Optimization Techniques","score":0.9053999781608582,"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":0.9053999781608582,"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"}}],"keywords":[{"id":"https://openalex.org/keywords/scalability","display_name":"Scalability","score":0.8191999793052673},{"id":"https://openalex.org/keywords/asynchronous-communication","display_name":"Asynchronous communication","score":0.6988999843597412},{"id":"https://openalex.org/keywords/robustness","display_name":"Robustness (evolution)","score":0.6111000180244446},{"id":"https://openalex.org/keywords/speedup","display_name":"Speedup","score":0.490200012922287},{"id":"https://openalex.org/keywords/reinforcement-learning","display_name":"Reinforcement learning","score":0.49000000953674316},{"id":"https://openalex.org/keywords/synchronization","display_name":"Synchronization (alternating current)","score":0.46209999918937683},{"id":"https://openalex.org/keywords/trajectory","display_name":"Trajectory","score":0.3991999924182892}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8309999704360962},{"id":"https://openalex.org/C48044578","wikidata":"https://www.wikidata.org/wiki/Q727490","display_name":"Scalability","level":2,"score":0.8191999793052673},{"id":"https://openalex.org/C151319957","wikidata":"https://www.wikidata.org/wiki/Q752739","display_name":"Asynchronous communication","level":2,"score":0.6988999843597412},{"id":"https://openalex.org/C120314980","wikidata":"https://www.wikidata.org/wiki/Q180634","display_name":"Distributed computing","level":1,"score":0.6480000019073486},{"id":"https://openalex.org/C63479239","wikidata":"https://www.wikidata.org/wiki/Q7353546","display_name":"Robustness (evolution)","level":3,"score":0.6111000180244446},{"id":"https://openalex.org/C68339613","wikidata":"https://www.wikidata.org/wiki/Q1549489","display_name":"Speedup","level":2,"score":0.490200012922287},{"id":"https://openalex.org/C97541855","wikidata":"https://www.wikidata.org/wiki/Q830687","display_name":"Reinforcement learning","level":2,"score":0.49000000953674316},{"id":"https://openalex.org/C2778562939","wikidata":"https://www.wikidata.org/wiki/Q1298791","display_name":"Synchronization (alternating current)","level":3,"score":0.46209999918937683},{"id":"https://openalex.org/C13662910","wikidata":"https://www.wikidata.org/wiki/Q193139","display_name":"Trajectory","level":2,"score":0.3991999924182892},{"id":"https://openalex.org/C2777303404","wikidata":"https://www.wikidata.org/wiki/Q759757","display_name":"Convergence (economics)","level":2,"score":0.38580000400543213},{"id":"https://openalex.org/C149810388","wikidata":"https://www.wikidata.org/wiki/Q5374873","display_name":"Emulation","level":2,"score":0.3822999894618988},{"id":"https://openalex.org/C2778156585","wikidata":"https://www.wikidata.org/wiki/Q174053","display_name":"Relay","level":3,"score":0.35010001063346863},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.3237999975681305},{"id":"https://openalex.org/C99844830","wikidata":"https://www.wikidata.org/wiki/Q102441924","display_name":"Scaling","level":2,"score":0.3230000138282776},{"id":"https://openalex.org/C82876162","wikidata":"https://www.wikidata.org/wiki/Q17096504","display_name":"Latency (audio)","level":2,"score":0.3190999925136566},{"id":"https://openalex.org/C138959212","wikidata":"https://www.wikidata.org/wiki/Q1806783","display_name":"Load balancing (electrical power)","level":3,"score":0.26910001039505005},{"id":"https://openalex.org/C79403827","wikidata":"https://www.wikidata.org/wiki/Q3988","display_name":"Real-time computing","level":1,"score":0.26330000162124634}],"mesh":[],"locations_count":3,"locations":[{"id":"doi:10.1145/3767295.3803580","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3767295.3803580","pdf_url":null,"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 21st European Conference on Computer Systems","raw_type":"proceedings-article"},{"id":"pmh:oai:arXiv.org:2510.12633","is_oa":true,"landing_page_url":"http://arxiv.org/abs/2510.12633","pdf_url":"https://arxiv.org/pdf/2510.12633","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":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2510.12633","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2510.12633","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":null,"license_id":null,"version":null,"is_accepted":false,"is_published":null,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"doi:10.1145/3767295.3803580","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3767295.3803580","pdf_url":null,"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 21st European Conference on Computer Systems","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Reinforcement":[0],"learning":[1],"(RL)":[2],"post-training":[3,125],"for":[4,17,79,196],"Large":[5],"Language":[6],"Models":[7],"(LLMs)":[8],"is":[9,32,77,99],"now":[10],"scaling":[11,102],"to":[12,20,51,159,210],"large":[13],"clusters":[14],"and":[15,64,83,112,122,153],"running":[16],"extended":[18],"durations":[19],"enhance":[21],"model":[22,71,220],"reasoning":[23],"performance.":[24],"However,":[25],"the":[26,62,80,161,167],"scalability":[27],"of":[28,86,141],"existing":[29],"RL":[30,39,48,91,124],"frameworks":[31],"limited,":[33],"as":[34,145],"extreme":[35],"long-tail":[36,177],"skewness":[37],"in":[38,90],"trajectory":[40,87,115],"generation":[41,88,185],"causes":[42],"severe":[43],"GPU":[44],"underutilization.":[45],"Current":[46],"asynchronous":[47,152],"systems":[49],"attempt":[50],"mitigate":[52],"this,":[53],"but":[54],"they":[55],"rely":[56],"on":[57,128,201],"global":[58,75,136],"weight":[59,155,163],"synchronization":[60,76],"between":[61],"actor":[63],"all":[65],"rollouts,":[66,183],"which":[67,110],"creates":[68],"a":[69,120,129,139,146,172,180,202],"rigid":[70],"update":[72],"schedule.":[73],"This":[74,150],"ill-suited":[78],"highly":[81],"skewed":[82],"evolving":[84],"distribution":[85],"latency":[89],"training,":[92],"crippling":[93],"training":[94,169,212],"efficiency.":[95],"Our":[96,199],"key":[97],"insight":[98],"that":[100,206],"efficient":[101],"requires":[103],"breaking":[104],"this":[105],"lockstep":[106],"through":[107],"trajectory-level":[108],"asynchrony,":[109],"generates":[111],"consumes":[113],"each":[114],"independently.":[116],"We":[117],"propose":[118],"Laminar,":[119],"scalable":[121],"robust":[123],"system":[126],"built":[127],"fully":[130,188],"decoupled":[131,189],"architecture.":[132],"First,":[133],"we":[134],"replace":[135],"updates":[137],"with":[138],"tier":[140],"relay":[142],"workers":[143],"acting":[144],"distributed":[147],"parameter":[148],"service.":[149],"enables":[151],"fine-grained":[154],"synchronization,":[156],"allowing":[157],"rollouts":[158],"pull":[160],"latest":[162],"anytime":[164],"without":[165],"stalling":[166],"actor's":[168],"loop.":[170],"Second,":[171],"dynamic":[173],"repack":[174],"mechanism":[175],"consolidates":[176],"trajectories":[178],"onto":[179],"few":[181],"dedicated":[182],"maximizing":[184],"throughput.":[186],"The":[187],"design":[190],"also":[191],"isolates":[192],"failures,":[193],"ensuring":[194],"robustness":[195],"long-running":[197],"jobs.":[198],"evaluation":[200],"1024-GPU":[203],"cluster":[204],"shows":[205],"Laminar":[207],"achieves":[208],"up":[209],"5.48$\\times$":[211],"throughput":[213],"speedup":[214],"over":[215],"state-of-the-art":[216],"systems,":[217],"while":[218],"reducing":[219],"convergence":[221],"time.":[222]},"counts_by_year":[],"updated_date":"2026-07-03T08:13:44.112507","created_date":"2025-10-17T00:00:00"}
