{"id":"https://openalex.org/W4412888864","doi":"https://doi.org/10.18653/v1/2025.findings-acl.205","title":"Do Large Language Models Perform Latent Multi-Hop Reasoning without Exploiting Shortcuts?","display_name":"Do Large Language Models Perform Latent Multi-Hop Reasoning without Exploiting Shortcuts?","publication_year":2025,"publication_date":"2025-01-01","ids":{"openalex":"https://openalex.org/W4412888864","doi":"https://doi.org/10.18653/v1/2025.findings-acl.205"},"language":"en","primary_location":{"id":"doi:10.18653/v1/2025.findings-acl.205","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-acl.205","pdf_url":"https://aclanthology.org/2025.findings-acl.205.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: ACL 2025","raw_type":"proceedings-article"},"type":"conference-paper","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://aclanthology.org/2025.findings-acl.205.pdf","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5023576577","display_name":"Sohee Yang","orcid":null},"institutions":[{"id":"https://openalex.org/I2801187875","display_name":"University College Lahore","ror":"https://ror.org/0572v5y16","country_code":"PK","type":"education","lineage":["https://openalex.org/I2801187875"]}],"countries":["PK"],"is_corresponding":false,"raw_author_name":"Sohee Yang","raw_affiliation_strings":["UCL"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"UCL","institution_ids":["https://openalex.org/I2801187875"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5019973471","display_name":"Nora Kassner","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Nora Kassner","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5040145850","display_name":"Elena Gribovskaya","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Elena Gribovskaya","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5101404695","display_name":"Sebastian Riedel","orcid":"https://orcid.org/0000-0002-3655-2486"},"institutions":[{"id":"https://openalex.org/I2801187875","display_name":"University College Lahore","ror":"https://ror.org/0572v5y16","country_code":"PK","type":"education","lineage":["https://openalex.org/I2801187875"]}],"countries":["PK"],"is_corresponding":false,"raw_author_name":"Sebastian Riedel","raw_affiliation_strings":["UCL"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"UCL","institution_ids":["https://openalex.org/I2801187875"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5094018628","display_name":"Mor Geva","orcid":null},"institutions":[{"id":"https://openalex.org/I1291425158","display_name":"Google (United States)","ror":"https://ror.org/00njsd438","country_code":"US","type":"company","lineage":["https://openalex.org/I1291425158","https://openalex.org/I4210128969"]},{"id":"https://openalex.org/I16391192","display_name":"Tel Aviv University","ror":"https://ror.org/04mhzgx49","country_code":"IL","type":"education","lineage":["https://openalex.org/I16391192"]}],"countries":["IL","US"],"is_corresponding":false,"raw_author_name":"Mor Geva","raw_affiliation_strings":["Google Research","Tel Aviv University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Google Research","institution_ids":["https://openalex.org/I1291425158"]},{"raw_affiliation_string":"Tel Aviv University","institution_ids":["https://openalex.org/I16391192"]}]}],"institutions":[],"countries_distinct_count":3,"institutions_distinct_count":3,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":2.9161,"has_fulltext":true,"cited_by_count":2,"citation_normalized_percentile":{"value":0.91019194,"is_in_top_1_percent":false,"is_in_top_10_percent":true},"cited_by_percentile_year":{"min":91,"max":97},"biblio":{"volume":null,"issue":null,"first_page":"3971","last_page":"3992"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10181","display_name":"Natural Language Processing Techniques","score":0.9944000244140625,"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/T10181","display_name":"Natural Language Processing Techniques","score":0.9944000244140625,"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":0.9836999773979187,"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/T12031","display_name":"Speech and dialogue systems","score":0.9147999882698059,"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":0.7996695041656494},{"id":"https://openalex.org/keywords/language-model","display_name":"Language model","score":0.5450239181518555},{"id":"https://openalex.org/keywords/hop","display_name":"Hop (telecommunications)","score":0.519332766532898},{"id":"https://openalex.org/keywords/natural-language-processing","display_name":"Natural language processing","score":0.39943528175354004},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.3772399425506592},{"id":"https://openalex.org/keywords/computer-network","display_name":"Computer network","score":0.12970438599586487}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7996695041656494},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.5450239181518555},{"id":"https://openalex.org/C25906391","wikidata":"https://www.wikidata.org/wiki/Q1432381","display_name":"Hop (telecommunications)","level":2,"score":0.519332766532898},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.39943528175354004},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.3772399425506592},{"id":"https://openalex.org/C31258907","wikidata":"https://www.wikidata.org/wiki/Q1301371","display_name":"Computer network","level":1,"score":0.12970438599586487}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.18653/v1/2025.findings-acl.205","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-acl.205","pdf_url":"https://aclanthology.org/2025.findings-acl.205.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: ACL 2025","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.18653/v1/2025.findings-acl.205","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-acl.205","pdf_url":"https://aclanthology.org/2025.findings-acl.205.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: ACL 2025","raw_type":"proceedings-article"},"sustainable_development_goals":[{"id":"https://metadata.un.org/sdg/10","score":0.4099999964237213,"display_name":"Reduced inequalities"}],"awards":[],"funders":[],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4412888864.pdf","grobid_xml":"https://content.openalex.org/works/W4412888864.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W4391375266","https://openalex.org/W2899084033","https://openalex.org/W2748952813","https://openalex.org/W2117210722","https://openalex.org/W2589759689","https://openalex.org/W4405141166","https://openalex.org/W1978191894","https://openalex.org/W2390279801","https://openalex.org/W2018045843","https://openalex.org/W3204019825"],"abstract_inverted_index":{"We":[0],"evaluate":[1],"how":[2],"well":[3],"Large":[4],"Language":[5],"Models":[6],"(LLMs)":[7],"latently":[8,179],"recall":[9,141,163],"and":[10,53,82,94,96,200],"compose":[11],"facts":[12,95],"to":[13,158,177],"answer":[14,55,68,83,189],"multi-hop":[15,125,206],"queries":[16,78,138,195],"like":[17],"\"In":[18],"the":[19,25,31,48,54,60,67,80,145,148,162,173,187,202],"year":[20],"Scarlett":[21],"Johansson":[22],"was":[23],"born,":[24],"Summer":[26],"Olympics":[27],"were":[28],"hosted":[29],"in":[30,36,59,194],"country":[32],"of\".One":[33],"major":[34],"challenge":[35],"such":[37],"evaluation":[38,113],"is":[39],"that":[40,120,183],"LLMs":[41,121],"may":[42],"have":[43,86],"developed":[44],"shortcuts":[45],"by":[46],"encountering":[47],"head":[49,81],"entity":[50,56],"\"Scarlett":[51],"Johansson\"":[52],"\"United":[57],"States\"":[58],"same":[61],"training":[62],"sequences":[63],"or":[64,106],"merely":[65],"guess":[66,104],"based":[69],"on":[70],"frequency-based":[71],"priors.To":[72],"prevent":[73],"shortcuts,":[74,130],"we":[75,110],"exclude":[76],"test":[77],"where":[79,101],"entities":[84],"might":[85,103],"co-appeared":[87],"during":[88,208],"training.Through":[89],"careful":[90],"selection":[91],"of":[92,99,136,142,164,175,186,204],"relations":[93],"systematic":[97],"removal":[98],"cases":[100],"models":[102,150,176],"answers":[105],"exploit":[107],"partial":[108],"matches,":[109],"construct":[111],"an":[112],"dataset":[114],"SOCRATES":[115],"(SHORTCUT-FREE":[116],"LATENT":[117],"REASONING).We":[118],"observe":[119],"demonstrate":[122],"promising":[123],"latent":[124,140,153,184,198,205],"reasoning":[126,207],"abilities":[127],"without":[128],"exploiting":[129],"but":[131,155],"only":[132],"for":[133,161],"certain":[134],"types":[135],"queries.For":[137],"requiring":[139],"countries":[143],"as":[144],"intermediate":[146,188],"answer,":[147],"best":[149],"achieve":[151],"80%":[152],"composability,":[154,199],"this":[156],"drops":[157],"just":[159],"5%":[160],"years.Comparisons":[165],"with":[166,196],"Chainof-Thought":[167],"highlight":[168],"a":[169],"significant":[170],"gap":[171],"between":[172],"ability":[174],"reason":[178],"versus":[180],"explicitly.Analysis":[181],"reveals":[182],"representations":[185],"are":[190],"constructed":[191],"more":[192],"often":[193],"higher":[197],"shows":[201],"emergence":[203],"pretraining.":[209]},"counts_by_year":[{"year":2026,"cited_by_count":1},{"year":2025,"cited_by_count":1}],"updated_date":"2026-08-18T07:49:30.821534","created_date":"2025-10-10T00:00:00"}