{"id":"https://openalex.org/W7225820708","doi":"https://doi.org/10.5281/zenodo.24144908","title":"Pipit: A Placement-First Inference Engine for Frontier Mixture-of-Experts Models on Consumer Hardware","display_name":"Pipit: A Placement-First Inference Engine for Frontier Mixture-of-Experts Models on Consumer Hardware","publication_year":2026,"publication_date":"2026-11-11","ids":{"openalex":"https://openalex.org/W7225820708","doi":"https://doi.org/10.5281/zenodo.24144908"},"language":"en","primary_location":{"id":"doi:10.5281/zenodo.24144908","doi":"https://doi.org/10.5281/zenodo.24144908","is_oa":true,"landing_page_url":"https://doi.org/10.5281/zenodo.24144908","pdf_url":null,"source":{"id":"https://openalex.org/S4306400562","display_name":"Zenodo (CERN European Organization for Nuclear Research)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I67311998","host_organization_name":"European Organization for Nuclear Research","host_organization_lineage":["https://openalex.org/I67311998"],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"},"volume":null,"type":"preprint","is_traditional":true,"indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.5281/zenodo.24144908","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5160888945","display_name":"Sayan Chowdhury","orcid":"https://orcid.org/0009-0004-8421-0087","observed_orcids":["https://orcid.org/0009-0004-8421-0087"]},"institutions":[],"countries":[],"is_corresponding":true,"raw_author_name":"Chowdhury, Sayan","raw_affiliation_strings":["Kalaris Labs"],"raw_orcid":"https://orcid.org/0009-0004-8421-0087","affiliations":[{"raw_affiliation_string":"Kalaris Labs","institution_ids":[]}]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":0,"corresponding_author_ids":["https://openalex.org/A5160888945"],"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/T10054","display_name":"Parallel Computing and Optimization Techniques","score":0.9541000127792358,"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":0.9541000127792358,"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/T10181","display_name":"Natural Language Processing Techniques","score":0.01489999983459711,"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/T10028","display_name":"Topic Modeling","score":0.006300000008195639,"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"}}],"keywords":[],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7785000205039978},{"id":"https://openalex.org/C57273362","wikidata":"https://www.wikidata.org/wiki/Q576722","display_name":"Decoding methods","level":2,"score":0.5270000100135803},{"id":"https://openalex.org/C46743427","wikidata":"https://www.wikidata.org/wiki/Q1341685","display_name":"Inference engine","level":3,"score":0.5105000138282776},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.4456999897956848},{"id":"https://openalex.org/C179518139","wikidata":"https://www.wikidata.org/wiki/Q5140297","display_name":"Coding (social sciences)","level":2,"score":0.41019999980926514},{"id":"https://openalex.org/C118505674","wikidata":"https://www.wikidata.org/wiki/Q42586063","display_name":"Encoder","level":2,"score":0.3587000072002411},{"id":"https://openalex.org/C9390403","wikidata":"https://www.wikidata.org/wiki/Q3966","display_name":"Computer hardware","level":1,"score":0.34130001068115234},{"id":"https://openalex.org/C77265313","wikidata":"https://www.wikidata.org/wiki/Q879844","display_name":"Rest (music)","level":2,"score":0.33009999990463257},{"id":"https://openalex.org/C188045654","wikidata":"https://www.wikidata.org/wiki/Q17148339","display_name":"Memory bandwidth","level":2,"score":0.31439998745918274},{"id":"https://openalex.org/C2780385302","wikidata":"https://www.wikidata.org/wiki/Q367158","display_name":"Protocol (science)","level":3,"score":0.314300000667572},{"id":"https://openalex.org/C111919701","wikidata":"https://www.wikidata.org/wiki/Q9135","display_name":"Operating system","level":1,"score":0.3095000088214874},{"id":"https://openalex.org/C113775141","wikidata":"https://www.wikidata.org/wiki/Q428691","display_name":"Computer engineering","level":1,"score":0.3050999939441681},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.2935999929904938},{"id":"https://openalex.org/C199360897","wikidata":"https://www.wikidata.org/wiki/Q9143","display_name":"Programming language","level":1,"score":0.2922999858856201},{"id":"https://openalex.org/C152124472","wikidata":"https://www.wikidata.org/wiki/Q1204361","display_name":"Redundancy (engineering)","level":2,"score":0.29089999198913574},{"id":"https://openalex.org/C149635348","wikidata":"https://www.wikidata.org/wiki/Q193040","display_name":"Embedded system","level":1,"score":0.29030001163482666},{"id":"https://openalex.org/C2778689934","wikidata":"https://www.wikidata.org/wiki/Q1313396","display_name":"Headline","level":2,"score":0.28189998865127563},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.27549999952316284},{"id":"https://openalex.org/C49154492","wikidata":"https://www.wikidata.org/wiki/Q5300","display_name":"Central processing unit","level":2,"score":0.26750001311302185},{"id":"https://openalex.org/C58328972","wikidata":"https://www.wikidata.org/wiki/Q184609","display_name":"Expert system","level":2,"score":0.2639000117778778},{"id":"https://openalex.org/C12186640","wikidata":"https://www.wikidata.org/wiki/Q6815743","display_name":"Memory model","level":3,"score":0.2540000081062317},{"id":"https://openalex.org/C79403827","wikidata":"https://www.wikidata.org/wiki/Q3988","display_name":"Real-time computing","level":1,"score":0.2522999942302704},{"id":"https://openalex.org/C185798385","wikidata":"https://www.wikidata.org/wiki/Q1161707","display_name":"Benchmark (surveying)","level":2,"score":0.25029999017715454}],"mesh":[],"locations_count":2,"locations":[{"id":"doi:10.5281/zenodo.24144908","doi":"https://doi.org/10.5281/zenodo.24144908","is_oa":true,"landing_page_url":"https://doi.org/10.5281/zenodo.24144908","pdf_url":null,"source":{"id":"https://openalex.org/S4306400562","display_name":"Zenodo (CERN European Organization for Nuclear Research)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I67311998","host_organization_name":"European Organization for Nuclear Research","host_organization_lineage":["https://openalex.org/I67311998"],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"},{"id":"doi:10.5281/zenodo.24144909","doi":"https://doi.org/10.5281/zenodo.24144909","is_oa":true,"landing_page_url":"https://doi.org/10.5281/zenodo.24144909","pdf_url":null,"source":{"id":"https://openalex.org/S4306400562","display_name":"Zenodo (CERN European Organization for Nuclear Research)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I67311998","host_organization_name":"European Organization for Nuclear Research","host_organization_lineage":["https://openalex.org/I67311998"],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"doi:10.5281/zenodo.24144908","doi":"https://doi.org/10.5281/zenodo.24144908","is_oa":true,"landing_page_url":"https://doi.org/10.5281/zenodo.24144908","pdf_url":null,"source":{"id":"https://openalex.org/S4306400562","display_name":"Zenodo (CERN European Organization for Nuclear Research)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I67311998","host_organization_name":"European Organization for Nuclear Research","host_organization_lineage":["https://openalex.org/I67311998"],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"},"sustainable_development_goals":[],"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":{"Mixture-of-experts":[0],"(MoE)":[1],"language":[2],"models":[3],"with":[4,28,111,181],"hundreds":[5],"of":[6,8,16,56,119],"billions":[7],"parameters":[9],"activate":[10],"only":[11],"a":[12,121,129,142,146,167,182],"few":[13],"per":[14,97],"cent":[15],"their":[17],"weights":[18,39],"for":[19,153,163],"each":[20,53],"token.":[21],"That":[22],"makes":[23],"them":[24],"runnable":[25],"on":[26,65,82,90,109],"machines":[27],"far":[29],"less":[30],"memory":[31,44],"than":[32],"the":[33,36,42,47,117,201],"model,":[34],"provided":[35],"right":[37,43,48],"expert":[38,144,148],"sit":[40],"in":[41,74,203],"tier":[45],"at":[46],"moment.":[49],"Today's":[50],"local":[51],"engines":[52],"solve":[54],"part":[55],"this.":[57],"Colibri":[58],"streams":[59],"experts":[60],"from":[61],"SSD":[62],"but":[63,85],"relies":[64],"general-purpose":[66],"kernels":[67,96,152],"and":[68,99,138,160,166],"has":[69],"repeatedly":[70],"patched":[71],"memory-safety":[72],"defects":[73],"its":[75,105,189],"loaders;":[76],"uzu":[77],"reaches":[78],"strong":[79],"speculative-decoding":[80],"speed":[81],"Apple":[83],"Silicon":[84],"is":[86],"not":[87],"yet":[88],"stable":[89],"other":[91],"operating":[92],"systems;":[93],"Magnitude":[94],"tunes":[95],"device":[98],"connects":[100],"to":[101],"coding":[102],"agents,":[103],"though":[104],"headline":[106],"speed-ups":[107],"rest":[108],"comparisons":[110],"mismatched":[112],"settings.":[113],"This":[114],"report":[115,193],"presents":[116],"design":[118],"Pipit,":[120],"memory-safe":[122],"Rust":[123],"engine":[124],"that":[125,184,206],"treats":[126],"decoding":[127],"as":[128],"bandwidth":[130],"problem":[131],"across":[132],"three":[133],"tiers":[134],"(GPU":[135],"memory,":[136],"RAM":[137],"NVMe).":[139],"It":[140],"combines":[141],"slot-slab":[143],"store,":[145],"learned":[147],"predictor,":[149],"bandwidth-saturating":[150],"CPU":[151],"six":[154],"instruction-set":[155],"levels,":[156],"union-aware":[157],"speculative":[158],"decoding,":[159],"validated":[161],"loaders":[162],"GGUF,":[164],"safetensors":[165],"new":[168],"streaming":[169],"format,":[170],"KXF.":[171],"We":[172],"pre-register":[173],"ten":[174],"falsifiable":[175],"hypotheses":[176],"against":[177,188,212],"same-machine":[178],"baselines,":[179],"together":[180],"protocol":[183],"scores":[185],"every":[186],"run":[187],"physical":[190],"roofline.":[191],"The":[192],"contains":[194],"no":[195],"measured":[196],"Pipit":[197],"results:":[198],"it":[199],"fixes":[200],"claims":[202],"advance,":[204],"so":[205],"forthcoming":[207],"results":[208],"can":[209],"be":[210],"judged":[211],"them.":[213]},"counts_by_year":[],"updated_date":"2026-11-12T12:28:57.059296","created_date":"2026-11-12T00:00:00"}