{"id":"https://openalex.org/W7134266197","doi":"https://doi.org/10.48550/arxiv.2603.05806","title":"MoE Lens -- An Expert Is All You Need","display_name":"MoE Lens -- An Expert Is All You Need","publication_year":2026,"publication_date":"2026-03-06","ids":{"openalex":"https://openalex.org/W7134266197","doi":"https://doi.org/10.48550/arxiv.2603.05806"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2603.05806","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2603.05806","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,"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":false,"raw_source_name":null,"raw_type":"Preprint"},"type":"preprint","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.48550/arxiv.2603.05806","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5067032498","display_name":"Marmik V. Chaudhari","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Chaudhari, Marmik","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5128431806","display_name":"Idhant Gulati","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Gulati, Idhant","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5028527445","display_name":"Nishkal Hundia","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Hundia, Nishkal","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5128571665","display_name":"Pranav Karra","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Karra, Pranav","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5126453516","display_name":"Shivam Raval","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Raval, Shivam","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/T11704","display_name":"Mobile Crowdsensing and Crowdsourcing","score":0.3562000095844269,"subfield":{"id":"https://openalex.org/subfields/1706","display_name":"Computer Science Applications"},"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/T11704","display_name":"Mobile Crowdsensing and Crowdsourcing","score":0.3562000095844269,"subfield":{"id":"https://openalex.org/subfields/1706","display_name":"Computer Science Applications"},"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/T11307","display_name":"Domain Adaptation and Few-Shot Learning","score":0.25839999318122864,"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/T13274","display_name":"Expert finding and Q&A systems","score":0.13439999520778656,"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"}}],"keywords":[{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.6715999841690063},{"id":"https://openalex.org/keywords/perplexity","display_name":"Perplexity","score":0.656499981880188},{"id":"https://openalex.org/keywords/similarity","display_name":"Similarity (geometry)","score":0.47040000557899475},{"id":"https://openalex.org/keywords/pruning","display_name":"Pruning","score":0.4415000081062317},{"id":"https://openalex.org/keywords/routing","display_name":"Routing (electronic design automation)","score":0.42100000381469727},{"id":"https://openalex.org/keywords/cosine-similarity","display_name":"Cosine similarity","score":0.4099999964237213},{"id":"https://openalex.org/keywords/subject-matter-expert","display_name":"Subject-matter expert","score":0.3684999942779541},{"id":"https://openalex.org/keywords/expert-system","display_name":"Expert system","score":0.35199999809265137}],"concepts":[{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.6715999841690063},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6708999872207642},{"id":"https://openalex.org/C100279451","wikidata":"https://www.wikidata.org/wiki/Q372193","display_name":"Perplexity","level":3,"score":0.656499981880188},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5975000262260437},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.5361999869346619},{"id":"https://openalex.org/C103278499","wikidata":"https://www.wikidata.org/wiki/Q254465","display_name":"Similarity (geometry)","level":3,"score":0.47040000557899475},{"id":"https://openalex.org/C108010975","wikidata":"https://www.wikidata.org/wiki/Q500094","display_name":"Pruning","level":2,"score":0.4415000081062317},{"id":"https://openalex.org/C74172769","wikidata":"https://www.wikidata.org/wiki/Q1446839","display_name":"Routing (electronic design automation)","level":2,"score":0.42100000381469727},{"id":"https://openalex.org/C2780762811","wikidata":"https://www.wikidata.org/wiki/Q1784941","display_name":"Cosine similarity","level":3,"score":0.4099999964237213},{"id":"https://openalex.org/C105002631","wikidata":"https://www.wikidata.org/wiki/Q4833645","display_name":"Subject-matter expert","level":3,"score":0.3684999942779541},{"id":"https://openalex.org/C124101348","wikidata":"https://www.wikidata.org/wiki/Q172491","display_name":"Data mining","level":1,"score":0.35850000381469727},{"id":"https://openalex.org/C58328972","wikidata":"https://www.wikidata.org/wiki/Q184609","display_name":"Expert system","level":2,"score":0.35199999809265137},{"id":"https://openalex.org/C3020580240","wikidata":"https://www.wikidata.org/wiki/Q663272","display_name":"Expert opinion","level":2,"score":0.3246000111103058},{"id":"https://openalex.org/C48145219","wikidata":"https://www.wikidata.org/wiki/Q1335365","display_name":"Security token","level":2,"score":0.2946999967098236},{"id":"https://openalex.org/C22367795","wikidata":"https://www.wikidata.org/wiki/Q7625208","display_name":"Structured prediction","level":2,"score":0.28839999437332153},{"id":"https://openalex.org/C119898033","wikidata":"https://www.wikidata.org/wiki/Q3433888","display_name":"Ensemble forecasting","level":2,"score":0.27869999408721924},{"id":"https://openalex.org/C2779227376","wikidata":"https://www.wikidata.org/wiki/Q6505497","display_name":"Layer (electronics)","level":2,"score":0.2775000035762787},{"id":"https://openalex.org/C70518039","wikidata":"https://www.wikidata.org/wiki/Q16000077","display_name":"Dimensionality reduction","level":2,"score":0.275299996137619},{"id":"https://openalex.org/C2777472644","wikidata":"https://www.wikidata.org/wiki/Q16968992","display_name":"Approximate inference","level":3,"score":0.2752000093460083},{"id":"https://openalex.org/C185798385","wikidata":"https://www.wikidata.org/wiki/Q1161707","display_name":"Benchmark (surveying)","level":2,"score":0.2685000002384186},{"id":"https://openalex.org/C114289077","wikidata":"https://www.wikidata.org/wiki/Q3284399","display_name":"Statistical model","level":2,"score":0.26179999113082886},{"id":"https://openalex.org/C2780451532","wikidata":"https://www.wikidata.org/wiki/Q759676","display_name":"Task (project management)","level":2,"score":0.2578999996185303},{"id":"https://openalex.org/C155846161","wikidata":"https://www.wikidata.org/wiki/Q1143367","display_name":"Graphical model","level":2,"score":0.2547000050544739},{"id":"https://openalex.org/C57273362","wikidata":"https://www.wikidata.org/wiki/Q576722","display_name":"Decoding methods","level":2,"score":0.25209999084472656}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2603.05806","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2603.05806","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,"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":"Preprint"}],"best_oa_location":{"id":"doi:10.48550/arxiv.2603.05806","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2603.05806","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,"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":false,"raw_source_name":null,"raw_type":"Preprint"},"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":{"Mixture":[0,170],"of":[1,25,34,60,107,120,171,197],"Experts":[2,172],"(MoE)":[3],"models":[4,173],"enable":[5],"parameter-efficient":[6],"scaling":[7],"through":[8,39,103,183],"sparse":[9],"expert":[10,35,53,161,185],"activations,":[11],"yet":[12],"optimizing":[13],"their":[14,26],"inference":[15,181],"and":[16,46,132,151,191],"memory":[17],"costs":[18],"remains":[19],"challenging":[20],"due":[21],"to":[22,55],"limited":[23],"understanding":[24],"specialization":[27,36],"behavior.":[28],"We":[29,98],"present":[30],"a":[31,83,104,159],"systematic":[32,105],"analysis":[33,59,106],"in":[37,200],"MoEs":[38],"two":[40],"complementary":[41],"approaches:":[42],"domain-specific":[43],"routing":[44,110,121],"patterns":[45],"an":[47],"early":[48],"decoding":[49],"framework":[50],"that":[51,65,113,169],"tracks":[52],"contributions":[54],"output":[56,91],"representations.":[57],"Our":[58,166],"the":[61,78,88,94,108],"DeepSeekMoE":[62],"model":[63,79,189],"reveals":[64],"despite":[66],"having":[67,144],"64":[68],"routed":[69],"experts":[70,116,134],"with":[71,87,141],"6":[72],"active":[73],"for":[74,135,180],"each":[75],"layer's":[76],"computation,":[77],"predominantly":[80],"relies":[81],"on":[82],"few":[84,115],"specialized":[85,125],"experts,":[86],"top-weighted":[89],"expert's":[90],"closely":[92],"approximating":[93],"full":[95],"ensemble":[96,133],"prediction.":[97],"quantitatively":[99],"validate":[100],"these":[101,201],"findings":[102],"token":[109],"distribution,":[111],"demonstrating":[112],"very":[114],"handle":[117],"over":[118],"50\\%":[119],"decisions":[122],"across":[123,162],"different":[124],"domains.":[126,165],"Hidden":[127],"state":[128],"similarity":[129,146],"between":[130],"single":[131,160],"every":[136],"layer":[137],"is":[138],"extremely":[139],"high,":[140],"some":[142],"layers":[143],"cosine":[145],"as":[147,149],"high":[148],"0.95":[150],"perplexity":[152],"increasing":[153],"by":[154],"only":[155],"5\\%":[156],"when":[157],"using":[158],"all":[163],"three":[164],"results":[167],"indicate":[168],"exhibit":[174],"concentrated":[175],"expertise":[176],"highlighting":[177],"potential":[178],"opportunities":[179],"optimization":[182],"targeted":[184],"pruning":[186],"while":[187],"maintaining":[188],"performance":[190],"opening":[192],"avenues":[193],"towards":[194],"studying":[195],"localization":[196],"learned":[198],"knowledge":[199],"models.":[202]},"counts_by_year":[],"updated_date":"2026-08-16T07:02:28.622633","created_date":"2026-03-10T00:00:00"}