{"id":"https://openalex.org/W4411112916","doi":"https://doi.org/10.18653/v1/2025.findings-naacl.107","title":"SimSMoE: Toward Efficient Training Mixture of Experts via Solving Representational Collapse","display_name":"SimSMoE: Toward Efficient Training Mixture of Experts via Solving Representational Collapse","publication_year":2025,"publication_date":"2025-01-01","ids":{"openalex":"https://openalex.org/W4411112916","doi":"https://doi.org/10.18653/v1/2025.findings-naacl.107"},"language":"en","primary_location":{"id":"doi:10.18653/v1/2025.findings-naacl.107","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-naacl.107","pdf_url":"https://aclanthology.org/2025.findings-naacl.107.pdf","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Findings of the Association for Computational Linguistics: NAACL 2025","raw_type":"proceedings-article"},"type":"article","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://aclanthology.org/2025.findings-naacl.107.pdf","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5035809896","display_name":"Giang Do","orcid":null},"institutions":[{"id":"https://openalex.org/I4210156221","display_name":"Allen Institute for Artificial Intelligence","ror":"https://ror.org/05w520734","country_code":"US","type":"nonprofit","lineage":["https://openalex.org/I4210156221"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Giang Do","raw_affiliation_strings":["Applied Artificial Intelligence Institute (A2I2) , Deakin University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Applied Artificial Intelligence Institute (A2I2) , Deakin University","institution_ids":["https://openalex.org/I4210156221"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5101871429","display_name":"Linh Le","orcid":"https://orcid.org/0000-0001-7169-9232"},"institutions":[{"id":"https://openalex.org/I4210156221","display_name":"Allen Institute for Artificial Intelligence","ror":"https://ror.org/05w520734","country_code":"US","type":"nonprofit","lineage":["https://openalex.org/I4210156221"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Hung Le","raw_affiliation_strings":["Applied Artificial Intelligence Institute (A2I2) , Deakin University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Applied Artificial Intelligence Institute (A2I2) , Deakin University","institution_ids":["https://openalex.org/I4210156221"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5085471517","display_name":"Truyen Tran","orcid":"https://orcid.org/0000-0001-6531-8907"},"institutions":[{"id":"https://openalex.org/I4210156221","display_name":"Allen Institute for Artificial Intelligence","ror":"https://ror.org/05w520734","country_code":"US","type":"nonprofit","lineage":["https://openalex.org/I4210156221"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Truyen Tran","raw_affiliation_strings":["Applied Artificial Intelligence Institute (A2I2) , Deakin University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Applied Artificial Intelligence Institute (A2I2) , Deakin University","institution_ids":["https://openalex.org/I4210156221"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":["https://openalex.org/I4210156221"],"apc_list":null,"apc_paid":null,"fwci":1.6508,"has_fulltext":true,"cited_by_count":1,"citation_normalized_percentile":{"value":0.86269868,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":91,"max":95},"biblio":{"volume":null,"issue":null,"first_page":"2012","last_page":"2025"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11512","display_name":"Anomaly Detection Techniques and Applications","score":0.7724999785423279,"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"}},"topics":[{"id":"https://openalex.org/T11512","display_name":"Anomaly Detection Techniques and Applications","score":0.7724999785423279,"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/T14222","display_name":"Knowledge Management and Technology","score":0.71670001745224,"subfield":{"id":"https://openalex.org/subfields/1803","display_name":"Management Science and Operations Research"},"field":{"id":"https://openalex.org/fields/18","display_name":"Decision Sciences"},"domain":{"id":"https://openalex.org/domains/2","display_name":"Social Sciences"}},{"id":"https://openalex.org/T10653","display_name":"Robot Manipulation and Learning","score":0.704200029373169,"subfield":{"id":"https://openalex.org/subfields/2207","display_name":"Control and Systems Engineering"},"field":{"id":"https://openalex.org/fields/22","display_name":"Engineering"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.6273828744888306},{"id":"https://openalex.org/keywords/training","display_name":"Training (meteorology)","score":0.5958899259567261},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.39252692461013794},{"id":"https://openalex.org/keywords/physics","display_name":"Physics","score":0.07824668288230896}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6273828744888306},{"id":"https://openalex.org/C2777211547","wikidata":"https://www.wikidata.org/wiki/Q17141490","display_name":"Training (meteorology)","level":2,"score":0.5958899259567261},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.39252692461013794},{"id":"https://openalex.org/C121332964","wikidata":"https://www.wikidata.org/wiki/Q413","display_name":"Physics","level":0,"score":0.07824668288230896},{"id":"https://openalex.org/C153294291","wikidata":"https://www.wikidata.org/wiki/Q25261","display_name":"Meteorology","level":1,"score":0.0}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.18653/v1/2025.findings-naacl.107","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-naacl.107","pdf_url":"https://aclanthology.org/2025.findings-naacl.107.pdf","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Findings of the Association for Computational Linguistics: NAACL 2025","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.18653/v1/2025.findings-naacl.107","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-naacl.107","pdf_url":"https://aclanthology.org/2025.findings-naacl.107.pdf","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Findings of the Association for Computational Linguistics: NAACL 2025","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4411112916.pdf","grobid_xml":"https://content.openalex.org/works/W4411112916.grobid-xml"},"referenced_works_count":1,"referenced_works":["https://openalex.org/W6601706463"],"related_works":["https://openalex.org/W4391375266","https://openalex.org/W2899084033","https://openalex.org/W2748952813","https://openalex.org/W230091440","https://openalex.org/W2390279801","https://openalex.org/W2233261550","https://openalex.org/W4391913857","https://openalex.org/W2358668433","https://openalex.org/W4396701345","https://openalex.org/W2810751659"],"abstract_inverted_index":{"Sparse":[0,55],"mixture":[1],"of":[2,22,26,57,63,106],"experts":[3,78],"(SMoE)":[4],"have":[5],"emerged":[6],"as":[7],"an":[8],"effective":[9,28],"approach":[10],"for":[11,93,126],"scaling":[12],"large":[13,90],"language":[14,91],"models":[15,92],"while":[16],"keeping":[17],"a":[18,60,69,80],"constant":[19],"computational":[20],"cost.Regardless":[21],"several":[23],"notable":[24],"successes":[25],"SMoE,":[27],"training":[29],"such":[30],"architecture":[31],"remains":[32],"elusive":[33],"due":[34],"to":[35,71,99],"the":[36,73,101,127],"representation":[37,74],"collapse":[38,75],"problem,":[39],"which":[40],"in":[41,124],"turn":[42],"harms":[43],"model":[44],"performance":[45,125],"and":[46,96,104,118],"causes":[47],"parameter":[48],"redundancy.In":[49],"this":[50],"work,":[51],"we":[52],"present":[53],"Similarity-based":[54],"Mixture":[56],"Experts":[58],"(SimSMoE),":[59],"novel":[61],"similarity":[62],"neural":[64],"network":[65],"algorithm,":[66],"that":[67,111],"guarantees":[68],"solution":[70],"address":[72],"issue":[76],"between":[77],"given":[79],"fixed":[81],"FLOPs":[82],"budget.We":[83],"conduct":[84],"extensive":[85],"empirical":[86],"evaluations":[87],"on":[88],"three":[89],"both":[94],"Pre-training":[95],"Fine-tuning":[97],"tasks":[98],"illustrate":[100],"efficacy,":[102],"robustness,":[103],"scalability":[105],"our":[107],"method.The":[108],"results":[109],"demonstrate":[110],"SimSMoE":[112],"significantly":[113],"enhances":[114],"existing":[115],"routing":[116,122],"policy":[117],"outperforms":[119],"other":[120],"SMoE":[121],"methods":[123],"tasks.Our":[128],"implementation":[129],"is":[130],"publicly":[131],"available":[132],"at":[133]},"counts_by_year":[{"year":2025,"cited_by_count":1}],"updated_date":"2026-06-26T08:34:08.712188","created_date":"2025-10-10T00:00:00"}