{"id":"https://openalex.org/W4285298148","doi":"https://doi.org/10.18653/v1/2022.acl-long.554","title":"Softmax Bottleneck Makes Language Models Unable to Represent Multi-mode Word Distributions","display_name":"Softmax Bottleneck Makes Language Models Unable to Represent Multi-mode Word Distributions","publication_year":2022,"publication_date":"2022-01-01","ids":{"openalex":"https://openalex.org/W4285298148","doi":"https://doi.org/10.18653/v1/2022.acl-long.554"},"language":"en","primary_location":{"id":"doi:10.18653/v1/2022.acl-long.554","doi":"https://doi.org/10.18653/v1/2022.acl-long.554","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2022.acl-long.554","pdf_url":"https://aclanthology.org/2022.acl-long.554.pdf","source":{"id":"https://openalex.org/S4393915740","display_name":"Proceedings of the conference - Association for Computational Linguistics. Meeting","issn_l":"0736-587X","issn":["0736-587X"],"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":["norway-1"],"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"conference"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"},"volume":null,"type":"conference-paper","is_traditional":true,"indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"hybrid","oa_url":"https://aclanthology.org/2022.acl-long.554.pdf","any_repository_has_fulltext":false},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5080115221","display_name":"Haw-Shiuan Chang","orcid":"https://orcid.org/0000-0003-4607-936X","observed_orcids":["https://orcid.org/0000-0003-4607-936X"]},"institutions":[{"id":"https://openalex.org/I24603500","display_name":"University of Massachusetts Amherst","ror":"https://ror.org/0072zz521","country_code":"US","type":"education","lineage":["https://openalex.org/I24603500"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Haw-Shiuan Chang","raw_affiliation_strings":["University of Massachusetts, Amherst 140 Governors Dr., Amherst, MA, USA 01003"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of Massachusetts, Amherst 140 Governors Dr., Amherst, MA, USA 01003","institution_ids":["https://openalex.org/I24603500"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5107835063","display_name":"Andrew McCallum","orcid":"https://orcid.org/0000-0003-2843-6992","observed_orcids":["https://orcid.org/0000-0003-2843-6992"]},"institutions":[{"id":"https://openalex.org/I24603500","display_name":"University of Massachusetts Amherst","ror":"https://ror.org/0072zz521","country_code":"US","type":"education","lineage":["https://openalex.org/I24603500"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Andrew McCallum","raw_affiliation_strings":["University of Massachusetts, Amherst 140 Governors Dr., Amherst, MA, USA 01003"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of Massachusetts, Amherst 140 Governors Dr., Amherst, MA, USA 01003","institution_ids":["https://openalex.org/I24603500"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":["https://openalex.org/I24603500"],"apc_list":null,"apc_paid":null,"fwci":1.4667,"has_fulltext":true,"cited_by_count":18,"citation_normalized_percentile":{"value":0.85296019,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":93,"max":98},"biblio":{"volume":null,"issue":null,"first_page":"8048","last_page":"8073"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10028","display_name":"Topic Modeling","score":0.9667999744415283,"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/T10028","display_name":"Topic Modeling","score":0.9667999744415283,"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/T10181","display_name":"Natural Language Processing Techniques","score":0.02459999918937683,"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/T11714","display_name":"Multimodal Machine Learning Applications","score":0.002199999988079071,"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"}}],"keywords":[{"id":"https://openalex.org/keywords/word-embeddings","display_name":"word embeddings","score":0.5040000081062317},{"id":"https://openalex.org/keywords/perplexity","display_name":"perplexity","score":0.625},{"id":"https://openalex.org/keywords/gpt-2","display_name":"GPT-2","score":0.8669999837875366},{"id":"https://openalex.org/keywords/bert","display_name":"BERT","score":0.9760000109672546}],"concepts":[{"id":"https://openalex.org/C100279451","wikidata":"https://www.wikidata.org/wiki/Q372193","display_name":"Perplexity","level":3,"score":0.9677886962890625},{"id":"https://openalex.org/C188441871","wikidata":"https://www.wikidata.org/wiki/Q7554146","display_name":"Softmax function","level":3,"score":0.9662227630615234},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7477627396583557},{"id":"https://openalex.org/C2780513914","wikidata":"https://www.wikidata.org/wiki/Q18210350","display_name":"Bottleneck","level":2,"score":0.6989023089408875},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.656679093837738},{"id":"https://openalex.org/C90805587","wikidata":"https://www.wikidata.org/wiki/Q10944557","display_name":"Word (group theory)","level":2,"score":0.6286632418632507},{"id":"https://openalex.org/C2777601683","wikidata":"https://www.wikidata.org/wiki/Q6499736","display_name":"Vocabulary","level":2,"score":0.6007969379425049},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5122994184494019},{"id":"https://openalex.org/C174348530","wikidata":"https://www.wikidata.org/wiki/Q188635","display_name":"Bridging (networking)","level":2,"score":0.47998282313346863},{"id":"https://openalex.org/C28490314","wikidata":"https://www.wikidata.org/wiki/Q189436","display_name":"Speech recognition","level":1,"score":0.43072953820228577},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.42539846897125244},{"id":"https://openalex.org/C50644808","wikidata":"https://www.wikidata.org/wiki/Q192776","display_name":"Artificial neural network","level":2,"score":0.32387131452560425},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.13464540243148804},{"id":"https://openalex.org/C41895202","wikidata":"https://www.wikidata.org/wiki/Q8162","display_name":"Linguistics","level":1,"score":0.08558624982833862},{"id":"https://openalex.org/C31258907","wikidata":"https://www.wikidata.org/wiki/Q1301371","display_name":"Computer network","level":1,"score":0.06933313608169556},{"id":"https://openalex.org/C138885662","wikidata":"https://www.wikidata.org/wiki/Q5891","display_name":"Philosophy","level":0,"score":0.0},{"id":"https://openalex.org/C149635348","wikidata":"https://www.wikidata.org/wiki/Q193040","display_name":"Embedded system","level":1,"score":0.0},{"id":"https://openalex.org/C2524010","wikidata":"https://www.wikidata.org/wiki/Q8087","display_name":"Geometry","level":1,"score":0.0}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.18653/v1/2022.acl-long.554","doi":"https://doi.org/10.18653/v1/2022.acl-long.554","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2022.acl-long.554","pdf_url":"https://aclanthology.org/2022.acl-long.554.pdf","source":{"id":"https://openalex.org/S4393915740","display_name":"Proceedings of the conference - Association for Computational Linguistics. Meeting","issn_l":"0736-587X","issn":["0736-587X"],"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":["norway-1"],"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"conference"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.18653/v1/2022.acl-long.554","doi":"https://doi.org/10.18653/v1/2022.acl-long.554","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2022.acl-long.554","pdf_url":"https://aclanthology.org/2022.acl-long.554.pdf","source":{"id":"https://openalex.org/S4393915740","display_name":"Proceedings of the conference - Association for Computational Linguistics. Meeting","issn_l":"0736-587X","issn":["0736-587X"],"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":["norway-1"],"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"conference"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"},"sustainable_development_goals":[],"study_designs":[],"awards":[{"id":"https://openalex.org/G1339228928","display_name":"RI: Medium: Tree-Structured Self-Supervised Modeling for Natural Language","funder_award_id":"1955567","funder_id":"https://openalex.org/F4320306076","funder_display_name":"National Science Foundation"},{"id":"https://openalex.org/G1667745721","display_name":"DMREF: Collaborative Research: The Synthesis Genome: Data Mining for Synthesis of New Materials","funder_award_id":"1534431","funder_id":"https://openalex.org/F4320306076","funder_display_name":"National Science Foundation"},{"id":"https://openalex.org/G2158607031","display_name":"RI: Medium: Extreme Clustering","funder_award_id":"1763618","funder_id":"https://openalex.org/F4320306076","funder_display_name":"National Science Foundation"}],"funders":[{"id":"https://openalex.org/F4320306076","display_name":"National Science Foundation","ror":"https://ror.org/021nxhr62"},{"id":"https://openalex.org/F4320308668","display_name":"University of Southern California","ror":"https://ror.org/03taz7m60"},{"id":"https://openalex.org/F4320315474","display_name":"Chan Zuckerberg Initiative","ror":"https://ror.org/02qenvm24"},{"id":"https://openalex.org/F4320337345","display_name":"Office of Naval Research","ror":"https://ror.org/00rk2pe57"},{"id":"https://openalex.org/F4320337367","display_name":"Division of Materials Research","ror":"https://ror.org/01pc7k308"}],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4285298148.pdf","grobid_xml":"https://content.openalex.org/works/W4285298148.grobid-xml"},"referenced_works_count":50,"referenced_works":["https://openalex.org/W1579853615","https://openalex.org/W2007548261","https://openalex.org/W2114082868","https://openalex.org/W2153579005","https://openalex.org/W2581624817","https://openalex.org/W2583105957","https://openalex.org/W2767321762","https://openalex.org/W2888799392","https://openalex.org/W2896232552","https://openalex.org/W2899663614","https://openalex.org/W2907252220","https://openalex.org/W2908510526","https://openalex.org/W2908854766","https://openalex.org/W2918117463","https://openalex.org/W2938704169","https://openalex.org/W2951210602","https://openalex.org/W2963127401","https://openalex.org/W2970184163","https://openalex.org/W2970466462","https://openalex.org/W2979826702","https://openalex.org/W2980360762","https://openalex.org/W3001279689","https://openalex.org/W3014104201","https://openalex.org/W3022096528","https://openalex.org/W3023509711","https://openalex.org/W3023533951","https://openalex.org/W3030163527","https://openalex.org/W3037932933","https://openalex.org/W3093524609","https://openalex.org/W3094152627","https://openalex.org/W3105816068","https://openalex.org/W3119519251","https://openalex.org/W3119866685","https://openalex.org/W3123673616","https://openalex.org/W3125516434","https://openalex.org/W3133264589","https://openalex.org/W3134454711","https://openalex.org/W3167346107","https://openalex.org/W3176904855","https://openalex.org/W3207166518","https://openalex.org/W3207548174","https://openalex.org/W3210904070","https://openalex.org/W3214373999","https://openalex.org/W4206297651","https://openalex.org/W4226287673","https://openalex.org/W4287894292","https://openalex.org/W4385574293","https://openalex.org/W4386506836","https://openalex.org/W4394667011","https://openalex.org/W7133224479"],"related_works":["https://openalex.org/W2767321762","https://openalex.org/W6958738254","https://openalex.org/W2804323070","https://openalex.org/W2918117463","https://openalex.org/W3096040360","https://openalex.org/W3176904855","https://openalex.org/W4394781992","https://openalex.org/W4385571072","https://openalex.org/W3022096528","https://openalex.org/W2739909019"],"abstract_inverted_index":{"Neural":[0],"language":[1],"models":[2],"(LMs)":[3],"such":[4],"as":[5],"GPT-2":[6,155],"estimate":[7],"the":[8,12,19,24,28,37,42,58,66,75,79,97,131,145,152,159,161,163],"probability":[9,54],"distribution":[10,25],"over":[11,18],"next":[13,81],"word":[14,89],"by":[15],"a":[16,32],"softmax":[17,21,113,118,127],"vocabulary.The":[20],"layer":[22],"produces":[23],"based":[26],"on":[27],"dot":[29],"products":[30],"of":[31,39,57,77,99,112,117,133,154],"single":[33,48,67],"hidden":[34,49,68],"state":[35,50,69],"and":[36,104,115,140,156],"embeddings":[38,76,90],"words":[40,82],"in":[41,151],"vocabulary.However,":[43],"we":[44,95],"discover":[45],"that":[46,142],"this":[47,93,100],"cannot":[51,71],"produce":[52],"all":[53,78],"distributions":[55],"regardless":[56],"LM":[59],"size":[60,64],"or":[61],"training":[62],"data":[63],"because":[65],"embedding":[70],"be":[72],"close":[73],"to":[74,124,129],"possible":[80],"simultaneously":[83],"when":[84],"there":[85],"are":[86],"other":[87],"interfering":[88],"between":[91],"them.In":[92],"work,":[94],"demonstrate":[96],"importance":[98],"limitation":[101],"both":[102],"theoretically":[103],"practically.Our":[105],"work":[106],"not":[107],"only":[108],"deepens":[109],"our":[110,138],"understanding":[111],"bottleneck":[114],"mixture":[116],"(MoS)":[119],"but":[120],"also":[121],"inspires":[122],"us":[123],"propose":[125],"multi-facet":[126],"(MFS)":[128],"address":[130],"limitations":[132],"MoS.Extensive":[134],"empirical":[135],"analyses":[136],"confirm":[137],"findings":[139],"show":[141],"against":[143],"MoS,":[144],"proposed":[146],"MFS":[147],"achieves":[148],"two-fold":[149],"improvements":[150],"perplexity":[153],"BERT.\"The":[157],"greater":[158,162],"ambiguity,":[160],"pleasure.\"-Milan":[164],"Kundera":[165]},"counts_by_year":[{"year":2026,"cited_by_count":4},{"year":2025,"cited_by_count":4},{"year":2024,"cited_by_count":6},{"year":2023,"cited_by_count":2},{"year":2022,"cited_by_count":2}],"updated_date":"2026-11-07T11:25:07.637180","created_date":"2022-07-14T00:00:00"}