{"id":"https://openalex.org/W4416245699","doi":"https://doi.org/10.48550/arxiv.2510.18921","title":"Benchmarking On-Device Machine Learning on Apple Silicon with MLX","display_name":"Benchmarking On-Device Machine Learning on Apple Silicon with MLX","publication_year":2025,"publication_date":"2025-10-21","ids":{"openalex":"https://openalex.org/W4416245699","doi":"https://doi.org/10.48550/arxiv.2510.18921"},"language":null,"primary_location":{"id":"pmh:oai:arXiv.org:2510.18921","is_oa":true,"landing_page_url":"http://arxiv.org/abs/2510.18921","pdf_url":"https://arxiv.org/pdf/2510.18921","source":{"id":"https://openalex.org/S4393918464","display_name":"ArXiv.org","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"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"},"type":"preprint","indexed_in":["arxiv","datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2510.18921","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5062369632","display_name":"Olasupo Ajayi","orcid":"https://orcid.org/0000-0001-6583-3749"},"institutions":[],"countries":[],"is_corresponding":true,"raw_author_name":"Ajayi, Oluwaseun A.","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5084090570","display_name":"Ogundepo Odunayo","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Odunayo, Ogundepo","raw_affiliation_strings":[],"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":2,"corresponding_author_ids":["https://openalex.org/A5062369632"],"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/T14347","display_name":"Big Data and Digital Economy","score":0.11800000071525574,"subfield":{"id":"https://openalex.org/subfields/1710","display_name":"Information Systems"},"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/T14347","display_name":"Big Data and Digital Economy","score":0.11800000071525574,"subfield":{"id":"https://openalex.org/subfields/1710","display_name":"Information Systems"},"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":0.10760000348091125,"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/T12072","display_name":"Machine Learning and Algorithms","score":0.09489999711513519,"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":[{"id":"https://openalex.org/keywords/porting","display_name":"Porting","score":0.7565000057220459},{"id":"https://openalex.org/keywords/implementation","display_name":"Implementation","score":0.7105000019073486},{"id":"https://openalex.org/keywords/benchmarking","display_name":"Benchmarking","score":0.6995000243186951},{"id":"https://openalex.org/keywords/transformer","display_name":"Transformer","score":0.6223000288009644},{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.6014999747276306},{"id":"https://openalex.org/keywords/architecture","display_name":"Architecture","score":0.42100000381469727},{"id":"https://openalex.org/keywords/deep-learning","display_name":"Deep learning","score":0.4081999957561493},{"id":"https://openalex.org/keywords/mobile-processor","display_name":"Mobile processor","score":0.38190001249313354}],"concepts":[{"id":"https://openalex.org/C106251023","wikidata":"https://www.wikidata.org/wiki/Q851989","display_name":"Porting","level":3,"score":0.7565000057220459},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7440000176429749},{"id":"https://openalex.org/C26713055","wikidata":"https://www.wikidata.org/wiki/Q245962","display_name":"Implementation","level":2,"score":0.7105000019073486},{"id":"https://openalex.org/C86251818","wikidata":"https://www.wikidata.org/wiki/Q816754","display_name":"Benchmarking","level":2,"score":0.6995000243186951},{"id":"https://openalex.org/C66322947","wikidata":"https://www.wikidata.org/wiki/Q11658","display_name":"Transformer","level":3,"score":0.6223000288009644},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.6014999747276306},{"id":"https://openalex.org/C118524514","wikidata":"https://www.wikidata.org/wiki/Q173212","display_name":"Computer architecture","level":1,"score":0.590399980545044},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.5062000155448914},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.4862000048160553},{"id":"https://openalex.org/C113775141","wikidata":"https://www.wikidata.org/wiki/Q428691","display_name":"Computer engineering","level":1,"score":0.44909998774528503},{"id":"https://openalex.org/C123657996","wikidata":"https://www.wikidata.org/wiki/Q12271","display_name":"Architecture","level":2,"score":0.42100000381469727},{"id":"https://openalex.org/C108583219","wikidata":"https://www.wikidata.org/wiki/Q197536","display_name":"Deep learning","level":2,"score":0.4081999957561493},{"id":"https://openalex.org/C1665295","wikidata":"https://www.wikidata.org/wiki/Q6887219","display_name":"Mobile processor","level":5,"score":0.38190001249313354},{"id":"https://openalex.org/C149635348","wikidata":"https://www.wikidata.org/wiki/Q193040","display_name":"Embedded system","level":1,"score":0.35350000858306885},{"id":"https://openalex.org/C45374587","wikidata":"https://www.wikidata.org/wiki/Q12525525","display_name":"Computation","level":2,"score":0.3467999994754791},{"id":"https://openalex.org/C115903868","wikidata":"https://www.wikidata.org/wiki/Q80993","display_name":"Software engineering","level":1,"score":0.34389999508857727},{"id":"https://openalex.org/C83283714","wikidata":"https://www.wikidata.org/wiki/Q121117","display_name":"Supercomputer","level":2,"score":0.31290000677108765},{"id":"https://openalex.org/C107598950","wikidata":"https://www.wikidata.org/wiki/Q259864","display_name":"Microarchitecture","level":2,"score":0.3028999865055084},{"id":"https://openalex.org/C2778119891","wikidata":"https://www.wikidata.org/wiki/Q477690","display_name":"CUDA","level":2,"score":0.2928999960422516},{"id":"https://openalex.org/C202491316","wikidata":"https://www.wikidata.org/wiki/Q272683","display_name":"Instruction set","level":2,"score":0.2824000120162964},{"id":"https://openalex.org/C2777115002","wikidata":"https://www.wikidata.org/wiki/Q7168246","display_name":"Performance prediction","level":2,"score":0.2612999975681305},{"id":"https://openalex.org/C2776221188","wikidata":"https://www.wikidata.org/wiki/Q21072556","display_name":"Design space exploration","level":2,"score":0.2554999887943268},{"id":"https://openalex.org/C2983448237","wikidata":"https://www.wikidata.org/wiki/Q1078276","display_name":"Language understanding","level":2,"score":0.2533000111579895}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2510.18921","is_oa":true,"landing_page_url":"http://arxiv.org/abs/2510.18921","pdf_url":"https://arxiv.org/pdf/2510.18921","source":{"id":"https://openalex.org/S4393918464","display_name":"ArXiv.org","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"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.2510.18921","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2510.18921","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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":null,"raw_source_name":null,"raw_type":"article"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2510.18921","is_oa":true,"landing_page_url":"http://arxiv.org/abs/2510.18921","pdf_url":"https://arxiv.org/pdf/2510.18921","source":{"id":"https://openalex.org/S4393918464","display_name":"ArXiv.org","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"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":[],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"The":[0,52,241],"recent":[1],"widespread":[2],"adoption":[3],"of":[4,22,46,49,87,93,100,148,155,200,213,222,229,238],"Large":[5],"Language":[6],"Models":[7],"(LLMs)":[8],"and":[9,32,41,79,127,134,146,207,216,249],"machine":[10,67],"learning":[11,68],"in":[12,18,105,125,132,246],"general":[13],"has":[14],"sparked":[15],"research":[16,113],"interest":[17],"exploring":[19],"the":[20,98,129,138,143,164,196,203,211,220],"possibilities":[21],"deploying":[23],"these":[24],"models":[25,157,173,181,201,228],"on":[26,71,90,182],"smaller":[27],"devices":[28,187],"such":[29],"as":[30],"laptops":[31],"mobile":[33],"phones.":[34],"This":[35,81],"creates":[36],"a":[37,63,84,116,234],"need":[38,165],"for":[39,66,166],"frameworks":[40],"approaches":[42],"that":[43],"are":[44],"capable":[45],"taking":[47],"advantage":[48],"on-device":[50,252],"hardware.":[51],"MLX":[53,106,126,139],"framework":[54,64,117],"was":[55],"created":[56],"to":[57,137,226],"address":[58],"this":[59,112],"need.":[60],"It":[61],"is":[62],"optimized":[65],"(ML)":[69],"computations":[70],"Apple":[72,149,184],"silicon":[73],"devices,":[74],"facilitating":[75],"easier":[76],"research,":[77],"experimentation,":[78],"prototyping.":[80],"paper":[82],"presents":[83],"performance":[85,99,199,212],"evaluation":[86],"MLX,":[88],"focusing":[89],"inference":[91,197],"latency":[92,198],"transformer":[94,102,123,156,180],"models.":[95],"We":[96,209],"compare":[97,195],"different":[101,122,179,230],"architecture":[103,145],"implementations":[104,124],"with":[107,202,219],"their":[108],"Pytorch":[109],"counterparts.":[110],"For":[111],"we":[114,194],"create":[115],"called":[118],"MLX-transformers":[119],"which":[120],"includes":[121],"downloads":[128],"model":[130],"checkpoints":[131],"pytorch":[133],"converts":[135],"it":[136],"format.":[140],"By":[141],"leveraging":[142],"advanced":[144],"capabilities":[147],"Silicon,":[150],"MLX-Transformers":[151],"enables":[152],"seamless":[153],"execution":[154],"directly":[158],"sourced":[159],"from":[160],"Hugging":[161],"Face,":[162],"eliminating":[163],"checkpoint":[167],"conversion":[168],"often":[169],"required":[170],"when":[171],"porting":[172],"between":[174],"frameworks.":[175],"Our":[176],"study":[177],"benchmarks":[178],"two":[183],"Silicon":[185],"macbook":[186],"against":[188],"an":[189],"NVIDIA":[190],"CUDA":[191],"GPU.":[192],"Specifically,":[193],"same":[204],"parameter":[205],"sizes":[206],"checkpoints.":[208],"evaluate":[210],"BERT,":[214],"RoBERTa,":[215],"XLM-RoBERTa":[217],"models,":[218],"intention":[221],"extending":[223],"future":[224],"work":[225],"include":[227],"modalities,":[231],"thus":[232],"providing":[233],"more":[235,250],"comprehensive":[236],"assessment":[237],"MLX's":[239,244],"capabilities.":[240],"results":[242],"highlight":[243],"potential":[245],"enabling":[247],"efficient":[248],"accessible":[251],"ML":[253],"applications":[254],"within":[255],"Apple's":[256],"ecosystem.":[257]},"counts_by_year":[],"updated_date":"2026-03-07T16:01:11.037858","created_date":"2025-10-24T00:00:00"}