{"id":"https://openalex.org/W4380996571","doi":"https://doi.org/10.1162/tacl_a_00562","title":"The Parallelism Tradeoff: Limitations of Log-Precision Transformers","display_name":"The Parallelism Tradeoff: Limitations of Log-Precision Transformers","publication_year":2023,"publication_date":"2023-01-01","ids":{"openalex":"https://openalex.org/W4380996571","doi":"https://doi.org/10.1162/tacl_a_00562"},"language":"en","primary_location":{"id":"doi:10.1162/tacl_a_00562","is_oa":true,"landing_page_url":"https://doi.org/10.1162/tacl_a_00562","pdf_url":"https://direct.mit.edu/tacl/article-pdf/doi/10.1162/tacl_a_00562/2131191/tacl_a_00562.pdf","source":{"id":"https://openalex.org/S2729999759","display_name":"Transactions of the Association for Computational Linguistics","issn_l":"2307-387X","issn":["2307-387X"],"is_oa":true,"is_in_doaj":true,"is_core":true,"host_organization":"https://openalex.org/P4310320244","host_organization_name":"Association for Computational Linguistics","host_organization_lineage":["https://openalex.org/P4310320244"],"host_organization_lineage_names":["Association for Computational Linguistics"],"type":"journal"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Transactions of the Association for Computational Linguistics","raw_type":"journal-article"},"type":"article","indexed_in":["crossref","doaj"],"open_access":{"is_oa":true,"oa_status":"diamond","oa_url":"https://direct.mit.edu/tacl/article-pdf/doi/10.1162/tacl_a_00562/2131191/tacl_a_00562.pdf","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5109946254","display_name":"William Merrill","orcid":null},"institutions":[{"id":"https://openalex.org/I57206974","display_name":"New York University","ror":"https://ror.org/0190ak572","country_code":"US","type":"education","lineage":["https://openalex.org/I57206974"]}],"countries":["US"],"is_corresponding":true,"raw_author_name":"William Merrill","raw_affiliation_strings":["Center for Data Science, New York University, New York, NY, USA. willm@nyu.edu"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Center for Data Science, New York University, New York, NY, USA. willm@nyu.edu","institution_ids":["https://openalex.org/I57206974"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5077726785","display_name":"Ashish Sabharwal","orcid":null},"institutions":[{"id":"https://openalex.org/I4210140341","display_name":"Allen Institute","ror":"https://ror.org/03cpe7c52","country_code":"US","type":"nonprofit","lineage":["https://openalex.org/I4210140341"]}],"countries":["US"],"is_corresponding":true,"raw_author_name":"Ashish Sabharwal","raw_affiliation_strings":["Allen Institute for AI, Seattle, WA, USA. ashishs@allenai.org"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Allen Institute for AI, Seattle, WA, USA. ashishs@allenai.org","institution_ids":["https://openalex.org/I4210140341"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":2,"corresponding_author_ids":["https://openalex.org/A5077726785","https://openalex.org/A5109946254"],"corresponding_institution_ids":["https://openalex.org/I4210140341","https://openalex.org/I57206974"],"apc_list":null,"apc_paid":null,"fwci":2.4501,"has_fulltext":true,"cited_by_count":15,"citation_normalized_percentile":{"value":0.91027822,"is_in_top_1_percent":false,"is_in_top_10_percent":true},"cited_by_percentile_year":{"min":90,"max":100},"biblio":{"volume":"11","issue":null,"first_page":"531","last_page":"545"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T12072","display_name":"Machine Learning and Algorithms","score":0.9997000098228455,"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/T12072","display_name":"Machine Learning and Algorithms","score":0.9997000098228455,"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/T10028","display_name":"Topic Modeling","score":0.994700014591217,"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":0.9916999936103821,"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/computer-science","display_name":"Computer science","score":0.7947264909744263},{"id":"https://openalex.org/keywords/transformer","display_name":"Transformer","score":0.6227985620498657},{"id":"https://openalex.org/keywords/logarithm","display_name":"Logarithm","score":0.551000714302063},{"id":"https://openalex.org/keywords/computational-complexity-theory","display_name":"Computational complexity theory","score":0.4772229492664337},{"id":"https://openalex.org/keywords/architecture","display_name":"Architecture","score":0.4736897945404053},{"id":"https://openalex.org/keywords/theoretical-computer-science","display_name":"Theoretical computer science","score":0.42806482315063477},{"id":"https://openalex.org/keywords/parallelizable-manifold","display_name":"Parallelizable manifold","score":0.4272735118865967},{"id":"https://openalex.org/keywords/algorithm","display_name":"Algorithm","score":0.4003632962703705},{"id":"https://openalex.org/keywords/mathematics","display_name":"Mathematics","score":0.1718321442604065},{"id":"https://openalex.org/keywords/electrical-engineering","display_name":"Electrical engineering","score":0.13168391585350037}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7947264909744263},{"id":"https://openalex.org/C66322947","wikidata":"https://www.wikidata.org/wiki/Q11658","display_name":"Transformer","level":3,"score":0.6227985620498657},{"id":"https://openalex.org/C39927690","wikidata":"https://www.wikidata.org/wiki/Q11197","display_name":"Logarithm","level":2,"score":0.551000714302063},{"id":"https://openalex.org/C179799912","wikidata":"https://www.wikidata.org/wiki/Q205084","display_name":"Computational complexity theory","level":2,"score":0.4772229492664337},{"id":"https://openalex.org/C123657996","wikidata":"https://www.wikidata.org/wiki/Q12271","display_name":"Architecture","level":2,"score":0.4736897945404053},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.42806482315063477},{"id":"https://openalex.org/C148047603","wikidata":"https://www.wikidata.org/wiki/Q1014612","display_name":"Parallelizable manifold","level":2,"score":0.4272735118865967},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.4003632962703705},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.1718321442604065},{"id":"https://openalex.org/C119599485","wikidata":"https://www.wikidata.org/wiki/Q43035","display_name":"Electrical engineering","level":1,"score":0.13168391585350037},{"id":"https://openalex.org/C153349607","wikidata":"https://www.wikidata.org/wiki/Q36649","display_name":"Visual arts","level":1,"score":0.0},{"id":"https://openalex.org/C134306372","wikidata":"https://www.wikidata.org/wiki/Q7754","display_name":"Mathematical analysis","level":1,"score":0.0},{"id":"https://openalex.org/C165801399","wikidata":"https://www.wikidata.org/wiki/Q25428","display_name":"Voltage","level":2,"score":0.0},{"id":"https://openalex.org/C127413603","wikidata":"https://www.wikidata.org/wiki/Q11023","display_name":"Engineering","level":0,"score":0.0},{"id":"https://openalex.org/C142362112","wikidata":"https://www.wikidata.org/wiki/Q735","display_name":"Art","level":0,"score":0.0}],"mesh":[],"locations_count":2,"locations":[{"id":"doi:10.1162/tacl_a_00562","is_oa":true,"landing_page_url":"https://doi.org/10.1162/tacl_a_00562","pdf_url":"https://direct.mit.edu/tacl/article-pdf/doi/10.1162/tacl_a_00562/2131191/tacl_a_00562.pdf","source":{"id":"https://openalex.org/S2729999759","display_name":"Transactions of the Association for Computational Linguistics","issn_l":"2307-387X","issn":["2307-387X"],"is_oa":true,"is_in_doaj":true,"is_core":true,"host_organization":"https://openalex.org/P4310320244","host_organization_name":"Association for Computational Linguistics","host_organization_lineage":["https://openalex.org/P4310320244"],"host_organization_lineage_names":["Association for Computational Linguistics"],"type":"journal"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Transactions of the Association for Computational Linguistics","raw_type":"journal-article"},{"id":"pmh:oai:doaj.org/article:29f71320a4e74b8c80c7db0491d6dd2c","is_oa":true,"landing_page_url":"https://doaj.org/article/29f71320a4e74b8c80c7db0491d6dd2c","pdf_url":null,"source":{"id":"https://openalex.org/S4306401280","display_name":"DOAJ (DOAJ: Directory of Open Access Journals)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by-sa","license_id":"https://openalex.org/licenses/cc-by-sa","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":"Transactions of the Association for Computational Linguistics, Vol 11, Pp 531-545 (2023)","raw_type":"article"}],"best_oa_location":{"id":"doi:10.1162/tacl_a_00562","is_oa":true,"landing_page_url":"https://doi.org/10.1162/tacl_a_00562","pdf_url":"https://direct.mit.edu/tacl/article-pdf/doi/10.1162/tacl_a_00562/2131191/tacl_a_00562.pdf","source":{"id":"https://openalex.org/S2729999759","display_name":"Transactions of the Association for Computational Linguistics","issn_l":"2307-387X","issn":["2307-387X"],"is_oa":true,"is_in_doaj":true,"is_core":true,"host_organization":"https://openalex.org/P4310320244","host_organization_name":"Association for Computational Linguistics","host_organization_lineage":["https://openalex.org/P4310320244"],"host_organization_lineage_names":["Association for Computational Linguistics"],"type":"journal"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Transactions of the Association for Computational Linguistics","raw_type":"journal-article"},"sustainable_development_goals":[],"awards":[{"id":"https://openalex.org/G5693414920","display_name":null,"funder_award_id":"1922658","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"}],"has_content":{"grobid_xml":false,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4380996571.pdf"},"referenced_works_count":30,"referenced_works":["https://openalex.org/W1479913975","https://openalex.org/W1593059705","https://openalex.org/W1999647826","https://openalex.org/W2006912660","https://openalex.org/W2039302045","https://openalex.org/W2049516232","https://openalex.org/W2088956149","https://openalex.org/W2125463200","https://openalex.org/W2415170901","https://openalex.org/W2612690371","https://openalex.org/W2750779823","https://openalex.org/W2908802752","https://openalex.org/W2952744660","https://openalex.org/W2997645120","https://openalex.org/W3014096773","https://openalex.org/W3214210379","https://openalex.org/W4210843940","https://openalex.org/W4212774754","https://openalex.org/W4224874866","https://openalex.org/W4230612566","https://openalex.org/W4288089799","https://openalex.org/W4290994975","https://openalex.org/W4298227433","https://openalex.org/W4385572831","https://openalex.org/W6739901393","https://openalex.org/W6753640285","https://openalex.org/W6755207826","https://openalex.org/W6769627184","https://openalex.org/W6772364761","https://openalex.org/W6778883912"],"related_works":["https://openalex.org/W4285069850","https://openalex.org/W2891970004","https://openalex.org/W1996916724","https://openalex.org/W4240963716","https://openalex.org/W1974634278","https://openalex.org/W1596637634","https://openalex.org/W2321918740","https://openalex.org/W2154273227","https://openalex.org/W1776186222","https://openalex.org/W2900794160"],"abstract_inverted_index":{"Abstract":[0],"Despite":[1],"their":[2,45],"omnipresence":[3],"in":[4,29,44,66,95],"modern":[5],"NLP,":[6],"characterizing":[7],"the":[8,30,59,108,117,130,155],"computational":[9],"power":[10,60],"of":[11,32,61,119,154],"transformer":[12,109,131],"neural":[13],"nets":[14,38],"remains":[15],"an":[16,96],"interesting":[17],"open":[18],"question.":[19],"We":[20,113],"prove":[21],"that":[22],"transformers":[23,62,85],"whose":[24,36],"arithmetic":[25],"precision":[26],"is":[27,140],"logarithmic":[28,82],"number":[31],"input":[33],"tokens":[34],"(and":[35],"feedforward":[37],"are":[39],"computable":[40],"using":[41,63,81],"space":[42],"linear":[43,90],"input)":[46],"can":[47,78],"be":[48,79],"simulated":[49],"by":[50],"constant-depth":[51],"logspace-uniform":[52],"threshold":[53],"circuits.":[54],"This":[55],"provides":[56],"insight":[57],"on":[58],"known":[64],"results":[65],"complexity":[67],"theory.":[68],"For":[69],"example,":[70],"if":[71],"L\u2260P":[72],"(i.e.,":[73],"not":[74],"all":[75],"poly-time":[76],"problems":[77],"solved":[80],"space),":[83],"then":[84],"cannot":[86],"even":[87],"accurately":[88],"solve":[89],"equalities":[91],"or":[92],"check":[93],"membership":[94],"arbitrary":[97],"context-free":[98],"grammar":[99],"with":[100],"empty":[101],"productions.":[102],"Our":[103],"result":[104],"intuitively":[105],"emerges":[106],"from":[107],"architecture\u2019s":[110],"high":[111],"parallelizability.":[112],"thus":[114],"speculatively":[115],"introduce":[116],"idea":[118],"a":[120,150],"fundamental":[121],"parallelism":[122,139],"tradeoff:":[123],"any":[124],"model":[125],"architecture":[126],"as":[127,129],"parallelizable":[128],"will":[132],"obey":[133],"limitations":[134],"similar":[135],"to":[136,142],"it.":[137],"Since":[138],"key":[141],"training":[143],"models":[144],"at":[145],"massive":[146],"scale,":[147],"this":[148],"suggests":[149],"potential":[151],"inherent":[152],"weakness":[153],"scaling":[156],"paradigm.":[157]},"counts_by_year":[{"year":2026,"cited_by_count":7},{"year":2025,"cited_by_count":6},{"year":2024,"cited_by_count":1},{"year":2023,"cited_by_count":1}],"updated_date":"2026-06-10T14:10:52.464848","created_date":"2025-10-10T00:00:00"}
