{"id":"https://openalex.org/W7164130631","doi":"https://doi.org/10.48550/arxiv.2606.09886","title":"SHAPE: Coalition-Aware Expert Pruning for Sparse Mixture-of-Experts LLMs","display_name":"SHAPE: Coalition-Aware Expert Pruning for Sparse Mixture-of-Experts LLMs","publication_year":2026,"publication_date":"2026-06-03","ids":{"openalex":"https://openalex.org/W7164130631","doi":"https://doi.org/10.48550/arxiv.2606.09886"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2606.09886","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.09886","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.2606.09886","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5138381287","display_name":"Yuhao Zhang","orcid":null},"institutions":[],"countries":[],"is_corresponding":true,"raw_author_name":"Zhang, Yuhao","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":0,"corresponding_author_ids":["https://openalex.org/A5138381287"],"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":0.9217000007629395,"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":0.9217000007629395,"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/T11612","display_name":"Stochastic Gradient Optimization Techniques","score":0.010700000450015068,"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/T10273","display_name":"IoT and Edge/Fog Computing","score":0.00800000037997961,"subfield":{"id":"https://openalex.org/subfields/1705","display_name":"Computer Networks and Communications"},"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/pruning","display_name":"Pruning","score":0.6965000033378601},{"id":"https://openalex.org/keywords/robustness","display_name":"Robustness (evolution)","score":0.5464000105857849},{"id":"https://openalex.org/keywords/expert-system","display_name":"Expert system","score":0.48159998655319214},{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.474700003862381},{"id":"https://openalex.org/keywords/trimming","display_name":"Trimming","score":0.39419999718666077},{"id":"https://openalex.org/keywords/set","display_name":"Set (abstract data type)","score":0.38679999113082886},{"id":"https://openalex.org/keywords/subject-matter-expert","display_name":"Subject-matter expert","score":0.3650999963283539},{"id":"https://openalex.org/keywords/expert-elicitation","display_name":"Expert elicitation","score":0.3107999861240387}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6984000205993652},{"id":"https://openalex.org/C108010975","wikidata":"https://www.wikidata.org/wiki/Q500094","display_name":"Pruning","level":2,"score":0.6965000033378601},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.6568999886512756},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5633999705314636},{"id":"https://openalex.org/C63479239","wikidata":"https://www.wikidata.org/wiki/Q7353546","display_name":"Robustness (evolution)","level":3,"score":0.5464000105857849},{"id":"https://openalex.org/C58328972","wikidata":"https://www.wikidata.org/wiki/Q184609","display_name":"Expert system","level":2,"score":0.48159998655319214},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.474700003862381},{"id":"https://openalex.org/C56951928","wikidata":"https://www.wikidata.org/wiki/Q3539213","display_name":"Trimming","level":2,"score":0.39419999718666077},{"id":"https://openalex.org/C177264268","wikidata":"https://www.wikidata.org/wiki/Q1514741","display_name":"Set (abstract data type)","level":2,"score":0.38679999113082886},{"id":"https://openalex.org/C105002631","wikidata":"https://www.wikidata.org/wiki/Q4833645","display_name":"Subject-matter expert","level":3,"score":0.3650999963283539},{"id":"https://openalex.org/C124101348","wikidata":"https://www.wikidata.org/wiki/Q172491","display_name":"Data mining","level":1,"score":0.323199987411499},{"id":"https://openalex.org/C72161134","wikidata":"https://www.wikidata.org/wiki/Q5421219","display_name":"Expert elicitation","level":2,"score":0.3107999861240387},{"id":"https://openalex.org/C81917197","wikidata":"https://www.wikidata.org/wiki/Q628760","display_name":"Selection (genetic algorithm)","level":2,"score":0.30390000343322754},{"id":"https://openalex.org/C116834253","wikidata":"https://www.wikidata.org/wiki/Q2039217","display_name":"Identification (biology)","level":2,"score":0.30219998955726624},{"id":"https://openalex.org/C175291020","wikidata":"https://www.wikidata.org/wiki/Q1156822","display_name":"Offset (computer science)","level":2,"score":0.29429998993873596},{"id":"https://openalex.org/C2776760102","wikidata":"https://www.wikidata.org/wiki/Q5139990","display_name":"Code (set theory)","level":3,"score":0.28859999775886536},{"id":"https://openalex.org/C159423971","wikidata":"https://www.wikidata.org/wiki/Q177251","display_name":"Associative property","level":2,"score":0.28450000286102295},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.2782999873161316},{"id":"https://openalex.org/C120936955","wikidata":"https://www.wikidata.org/wiki/Q2155640","display_name":"Empirical research","level":2,"score":0.27390000224113464},{"id":"https://openalex.org/C164226766","wikidata":"https://www.wikidata.org/wiki/Q7293202","display_name":"Rank (graph theory)","level":2,"score":0.2606000006198883},{"id":"https://openalex.org/C113174947","wikidata":"https://www.wikidata.org/wiki/Q2859736","display_name":"Tree (set theory)","level":2,"score":0.2556999921798706},{"id":"https://openalex.org/C2779530757","wikidata":"https://www.wikidata.org/wiki/Q1207505","display_name":"Quality (philosophy)","level":2,"score":0.2549000084400177}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2606.09886","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.09886","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.2606.09886","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.09886","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":[{"display_name":"Partnerships for the goals","id":"https://metadata.un.org/sdg/17","score":0.5793706774711609}],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Sparse":[0],"Mixture-of-Experts":[1],"(MoE)":[2],"large":[3],"language":[4],"models":[5,72],"achieve":[6],"strong":[7],"quality":[8],"with":[9],"low":[10],"per-token":[11],"compute,":[12],"yet":[13],"their":[14],"deployment":[15],"is":[16,36,52,209],"often":[17],"limited":[18],"by":[19],"the":[20,23,104,139],"memory":[21,204],"wall:":[22],"full":[24],"expert":[25,61,74,93,141,192],"pool":[26],"must":[27],"remain":[28],"resident":[29],"to":[30,154],"support":[31],"token-dependent":[32],"routing.":[33],"Expert":[34],"pruning":[35,68,125,183,193],"a":[37,66,81,96,123,130,156],"direct":[38],"remedy,":[39],"but":[40],"prior":[41],"criteria":[42],"typically":[43],"score":[44],"experts":[45,107],"independently":[46],"and":[47,90,168,181,190,197],"overlook":[48],"that":[49,70,108,134,174],"MoE":[50,120,164],"inference":[51],"inherently":[53],"\\emph{coalitional},":[54],"where":[55],"outputs":[56],"arise":[57],"from":[58],"routed":[59],"top-$k$":[60,101],"combinations.":[62],"We":[63],"propose":[64],"\\textbf{SHAPE},":[65],"task-driven":[67],"framework":[69],"explicitly":[71],"\\emph{intra-layer}":[73],"cooperation.":[75],"SHAPE":[76,127,175],"formulates":[77],"routing":[78],"traces":[79],"on":[80,161],"small":[82],"calibration":[83],"set":[84],"as":[85],"an":[86,144],"empirical":[87],"cooperative":[88],"game":[89],"assigns":[91],"interaction-aware":[92],"values":[94],"via":[95],"Shapley-style":[97],"attribution":[98],"over":[99,179],"observed":[100],"coalitions,":[102],"enabling":[103],"identification":[105],"of":[106,147],"are":[109],"essential":[110],"for":[111],"high-utility":[112],"collaborations":[113],"rather":[114],"than":[115],"merely":[116],"frequent.":[117],"To":[118],"preserve":[119],"topology":[121],"under":[122,188],"global":[124,180],"budget,":[126],"further":[128],"introduces":[129],"\\emph{quality-coverage}":[131],"selection":[132],"rule":[133],"retains,":[135],"in":[136,201],"each":[137],"layer,":[138],"minimal":[140],"subset":[142],"covering":[143],"$\u03b1$":[145],"fraction":[146],"non-negative":[148],"Shapley":[149],"mass,":[150],"while":[151],"using":[152],"bisection":[153],"match":[155],"target":[157],"keep":[158],"rate.":[159],"Experiments":[160],"three":[162],"modern":[163],"backbones":[165],"(Qwen3-30B-A3B,":[166],"GPT-OSS-20B,":[167],"DeepSeek-V2-Lite)":[169],"across":[170],"diverse":[171],"benchmarks":[172],"show":[173],"consistently":[176],"improves":[177],"robustness":[178],"layer-wise":[182],"variants,":[184],"maintaining":[185],"competitive":[186],"accuracy":[187],"20\\%":[189],"40\\%":[191],"without":[194],"additional":[195],"training":[196],"delivering":[198],"clear":[199],"reductions":[200],"peak":[202],"GPU":[203],"footprint.":[205],"The":[206],"open-source":[207],"code":[208],"available":[210],"at":[211],"https://github.com/Alizen-1009/Shapley-Moe.":[212]},"counts_by_year":[],"updated_date":"2026-07-28T07:46:37.118299","created_date":"2026-06-11T00:00:00"}