{"id":"https://openalex.org/W7167477223","doi":"https://doi.org/10.1162/tacl.a.745","title":"From Benchmarks to Skills: Low-Rank Factors for LLM Evaluation","display_name":"From Benchmarks to Skills: Low-Rank Factors for LLM Evaluation","publication_year":2026,"publication_date":"2026-01-01","ids":{"openalex":"https://openalex.org/W7167477223","doi":"https://doi.org/10.1162/tacl.a.745"},"language":"en","primary_location":{"id":"doi:10.1162/tacl.a.745","doi":"https://doi.org/10.1162/tacl.a.745","is_oa":true,"landing_page_url":"https://doi.org/10.1162/tacl.a.745","pdf_url":null,"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,"listed_in":["ccf-b","cwts-core","doaj","dongbi-b","jufo-1","norway-2","poland-20","russia-white-list-1"],"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"},"volume":null,"type":"article","indexed_in":["arxiv","crossref","datacite","doaj"],"open_access":{"is_oa":true,"oa_status":"diamond","oa_url":"https://doi.org/10.1162/tacl.a.745","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5092998404","display_name":"Aviya Maimon","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I13955877","display_name":"Bar-Ilan University","ror":"https://ror.org/03kgsv495","country_code":"IL","type":"education","lineage":["https://openalex.org/I13955877"]}],"countries":["IL"],"is_corresponding":false,"raw_author_name":"Aviya Maimon","raw_affiliation_strings":["Bar-Ilan University, aviyamn@gmail.com","OriginAI"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Bar-Ilan University, aviyamn@gmail.com","institution_ids":["https://openalex.org/I13955877"]},{"raw_affiliation_string":"OriginAI","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5140113035","display_name":"Amir David Nisan Cohen","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Amir David Nisan Cohen","raw_affiliation_strings":["OriginAI, amirdnc@gmail.com"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"OriginAI, amirdnc@gmail.com","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5031360025","display_name":"Gal Vishne","orcid":"https://orcid.org/0000-0001-7906-2543","observed_orcids":["https://orcid.org/0000-0001-7906-2543"]},"institutions":[{"id":"https://openalex.org/I78577930","display_name":"Columbia University","ror":"https://ror.org/00hj8s172","country_code":"US","type":"education","lineage":["https://openalex.org/I78577930"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Gal Vishne","raw_affiliation_strings":["Data Science Institute, Columbia University, gal.vishne@gmail.com"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Data Science Institute, Columbia University, gal.vishne@gmail.com","institution_ids":["https://openalex.org/I78577930"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5072633418","display_name":"Shauli Ravfogel","orcid":null,"observed_orcids":[]},"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":false,"raw_author_name":"Shauli Ravfogel","raw_affiliation_strings":["Center for Data Science, New York University, shauli.ravfogel@gmail.com"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Center for Data Science, New York University, shauli.ravfogel@gmail.com","institution_ids":["https://openalex.org/I57206974"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5063283689","display_name":"Reut Tsarfaty","orcid":"https://orcid.org/0000-0001-6406-6568","observed_orcids":["https://orcid.org/0000-0001-6406-6568"]},"institutions":[{"id":"https://openalex.org/I13955877","display_name":"Bar-Ilan University","ror":"https://ror.org/03kgsv495","country_code":"IL","type":"education","lineage":["https://openalex.org/I13955877"]}],"countries":["IL"],"is_corresponding":false,"raw_author_name":"Reut Tsarfaty","raw_affiliation_strings":["Bar-Ilan University, reut.tsarfaty@gmail.com"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Bar-Ilan University, reut.tsarfaty@gmail.com","institution_ids":["https://openalex.org/I13955877"]}]}],"institutions":[],"countries_distinct_count":2,"institutions_distinct_count":3,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":true,"cited_by_count":0,"citation_normalized_percentile":{"value":0.45884241,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":"14","issue":null,"first_page":"1639","last_page":"1653"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10028","display_name":"Topic Modeling","score":0.9535999894142151,"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":0.9535999894142151,"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":0.02889999933540821,"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":0.002400000113993883,"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/llm-evaluation","display_name":"LLM evaluation","score":0.9300000071525574},{"id":"https://openalex.org/keywords/low-rank-structure","display_name":"low-rank structure","score":0.8220000267028809},{"id":"https://openalex.org/keywords/factor-analysis","display_name":"factor analysis","score":0.7590000033378601},{"id":"https://openalex.org/keywords/model-selection","display_name":"model selection","score":0.652999997138977},{"id":"https://openalex.org/keywords/interpretability","display_name":"interpretability","score":0.5370000004768372}],"concepts":[{"id":"https://openalex.org/C86251818","wikidata":"https://www.wikidata.org/wiki/Q816754","display_name":"Benchmarking","level":2,"score":0.9052000045776367},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8184999823570251},{"id":"https://openalex.org/C185798385","wikidata":"https://www.wikidata.org/wiki/Q1161707","display_name":"Benchmark (surveying)","level":2,"score":0.7087000012397766},{"id":"https://openalex.org/C152124472","wikidata":"https://www.wikidata.org/wiki/Q1204361","display_name":"Redundancy (engineering)","level":2,"score":0.5849999785423279},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.5626999735832214},{"id":"https://openalex.org/C4679612","wikidata":"https://www.wikidata.org/wiki/Q866298","display_name":"Aggregate (composite)","level":2,"score":0.5105000138282776},{"id":"https://openalex.org/C187191949","wikidata":"https://www.wikidata.org/wiki/Q1138496","display_name":"Profiling (computer programming)","level":2,"score":0.4203000068664551},{"id":"https://openalex.org/C2522767166","wikidata":"https://www.wikidata.org/wiki/Q2374463","display_name":"Data science","level":1,"score":0.4171000123023987},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.39989998936653137},{"id":"https://openalex.org/C2780451532","wikidata":"https://www.wikidata.org/wiki/Q759676","display_name":"Task (project management)","level":2,"score":0.34200000762939453},{"id":"https://openalex.org/C124101348","wikidata":"https://www.wikidata.org/wiki/Q172491","display_name":"Data mining","level":1,"score":0.33169999718666077},{"id":"https://openalex.org/C2778058735","wikidata":"https://www.wikidata.org/wiki/Q4692253","display_name":"Aggregate data","level":2,"score":0.31940001249313354},{"id":"https://openalex.org/C2781039887","wikidata":"https://www.wikidata.org/wiki/Q1391724","display_name":"Factor (programming language)","level":2,"score":0.2994999885559082},{"id":"https://openalex.org/C2164484","wikidata":"https://www.wikidata.org/wiki/Q5170150","display_name":"Core (optical fiber)","level":2,"score":0.2937999963760376},{"id":"https://openalex.org/C133462117","wikidata":"https://www.wikidata.org/wiki/Q4929239","display_name":"Data collection","level":2,"score":0.26260000467300415},{"id":"https://openalex.org/C51632099","wikidata":"https://www.wikidata.org/wiki/Q3985153","display_name":"Training set","level":2,"score":0.26190000772476196}],"mesh":[],"locations_count":3,"locations":[{"id":"doi:10.1162/tacl.a.745","doi":"https://doi.org/10.1162/tacl.a.745","is_oa":true,"landing_page_url":"https://doi.org/10.1162/tacl.a.745","pdf_url":null,"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,"listed_in":["ccf-b","cwts-core","doaj","dongbi-b","jufo-1","norway-2","poland-20","russia-white-list-1"],"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:arXiv.org:2507.20208","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2507.20208","pdf_url":"https://arxiv.org/pdf/2507.20208","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,"listed_in":[],"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":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2507.20208","doi":"https://doi.org/10.48550/arxiv.2507.20208","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2507.20208","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,"listed_in":[],"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"}],"best_oa_location":{"id":"doi:10.1162/tacl.a.745","doi":"https://doi.org/10.1162/tacl.a.745","is_oa":true,"landing_page_url":"https://doi.org/10.1162/tacl.a.745","pdf_url":null,"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,"listed_in":["ccf-b","cwts-core","doaj","dongbi-b","jufo-1","norway-2","poland-20","russia-white-list-1"],"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":[],"study_designs":[],"awards":[],"funders":[{"id":"https://openalex.org/F4320321950","display_name":"Hebrew University of Jerusalem","ror":"https://ror.org/03qxff017"},{"id":"https://openalex.org/F4320322252","display_name":"Israel Science Foundation","ror":"https://ror.org/04sazxf24"},{"id":"https://openalex.org/F4320335375","display_name":"Planning and Budgeting Committee of the Council for Higher Education of Israel","ror":null}],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":20,"referenced_works":["https://openalex.org/W1964218847","https://openalex.org/W1977775666","https://openalex.org/W2057069782","https://openalex.org/W2078483536","https://openalex.org/W2087399108","https://openalex.org/W2152918935","https://openalex.org/W2159306398","https://openalex.org/W2160172778","https://openalex.org/W2487770199","https://openalex.org/W2606974598","https://openalex.org/W2888482885","https://openalex.org/W2890894339","https://openalex.org/W2920807444","https://openalex.org/W3190752170","https://openalex.org/W4231950914","https://openalex.org/W4297625931","https://openalex.org/W4386576685","https://openalex.org/W4392637287","https://openalex.org/W4401982386","https://openalex.org/W7133239970"],"related_works":["https://openalex.org/W7223417875","https://openalex.org/W4416036262","https://openalex.org/W7162102073","https://openalex.org/W4391335357","https://openalex.org/W7214034687","https://openalex.org/W6930876559","https://openalex.org/W7114529973","https://openalex.org/W7129079999","https://openalex.org/W7162801773","https://openalex.org/W7129097951"],"abstract_inverted_index":{"Abstract":[0],"Current":[1],"evaluations":[2],"of":[3,14,64,80,92,100,105,145,175,196],"large":[4],"language":[5],"models":[6,170,180],"(LLMs)":[7],"rely":[8],"heavily":[9],"on":[10,17,60],"a":[11,42,61,76,97,172,189,197],"growing":[12],"collection":[13],"benchmarks":[15,83,126],"and":[16,30,122,160,177,201,205,210],"aggregate":[18,199],"benchmark":[19,51],"scores,":[20],"yet":[21],"it":[22],"remains":[23],"unclear":[24],"what":[25,31],"this":[26,149],"comparison":[27],"actually":[28],"captures,":[29],"these":[32,137],"scores":[33],"reveal":[34],"about":[35],"models\u2019":[36],"underlying":[37],"capabilities.":[38,214],"Here,":[39],"we":[40,70,152],"propose":[41],"new":[43,169],"paradigm":[44],"for":[45,157,208],"LLM":[46,146,158,212],"evaluation,":[47],"by":[48],"asking":[49],"whether":[50],"performance":[52,78],"reflects":[53],"many":[54,125],"independent":[55],"abilities,":[56],"or":[57],"rather,":[58],"relies":[59],"small":[62,98,173],"number":[63,99],"shared":[65],"dimensions.":[66],"To":[67],"answer":[68],"this,":[69],"apply":[71],"Factor":[72],"Analysis":[73],"(FA)":[74],"to":[75,128,141,192],"massive":[77],"matrix":[79],"LLMs":[81],"versus":[82],"(60":[84],"\u00d7":[85],"44)":[86],"revealing":[87],"an":[88,203],"intrinsically":[89],"low-rank":[90,114],"structure":[91,107],"that":[93,136],"matrix.":[94],"That":[95],"is,":[96],"latent":[101,138,150],"factors":[102,139],"captures":[103],"most":[104],"the":[106,109,193],"in":[108],"full":[110],"task":[111],"space.":[112],"This":[113],"geometry":[115],"reveals":[116],"substantial":[117],"redundancy":[118],"across":[119],"existing":[120],"tasks":[121],"explains":[123],"why":[124],"appear":[127],"be":[129],"measuring":[130],"overlapping":[131],"abilities.":[132],"We":[133],"further":[134],"show":[135],"correspond":[140],"coherent,":[142],"skill-like,":[143],"dimensions":[144],"behavior.":[147],"Leveraging":[148],"skill-space,":[151],"deliver":[153],"three":[154],"practical":[155,206],"tools":[156],"evaluation":[159],"downstream":[161],"users:":[162],"(i)":[163],"identifying":[164],"redundant":[165],"tasks,":[166,176],"(ii)":[167],"profiling":[168],"using":[171],"subset":[174],"(iii)":[178],"selecting":[179],"aligned":[181],"with":[182],"desired":[183],"skill":[184],"profiles.":[185],"Our":[186],"method":[187],"provides":[188],"solid":[190],"alternative":[191],"de-facto":[194],"standard":[195],"single":[198],"score,":[200],"establishes":[202],"interpretable":[204],"framework":[207],"understanding":[209],"benchmarking":[211],"core":[213]},"counts_by_year":[],"updated_date":"2026-10-31T12:09:11.096373","created_date":"2026-07-07T00:00:00"}