{"id":"https://openalex.org/W4415329122","doi":"https://doi.org/10.48550/arxiv.2505.15508","title":"Multilingual Test-Time Scaling via Initial Thought Transfer","display_name":"Multilingual Test-Time Scaling via Initial Thought Transfer","publication_year":2025,"publication_date":"2025-05-21","ids":{"openalex":"https://openalex.org/W4415329122","doi":"https://doi.org/10.48550/arxiv.2505.15508"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2505.15508","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2505.15508","pdf_url":"https://arxiv.org/pdf/2505.15508","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"},"volume":null,"type":"preprint","indexed_in":["arxiv","datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2505.15508","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5107956331","display_name":"Prasoon Bajpai","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I68891433","display_name":"Indian Institute of Technology Delhi","ror":"https://ror.org/049tgcd06","country_code":"IN","type":"education","lineage":["https://openalex.org/I68891433"]}],"countries":["IN"],"is_corresponding":false,"raw_author_name":"Bajpai, Prasoon","raw_affiliation_strings":["IIT Delhi"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"IIT Delhi","institution_ids":["https://openalex.org/I68891433"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5046521217","display_name":"Tanmoy Chakraborty","orcid":"https://orcid.org/0000-0002-0210-0369","observed_orcids":["https://orcid.org/0000-0002-0210-0369"]},"institutions":[{"id":"https://openalex.org/I68891433","display_name":"Indian Institute of Technology Delhi","ror":"https://ror.org/049tgcd06","country_code":"IN","type":"education","lineage":["https://openalex.org/I68891433"]}],"countries":["IN"],"is_corresponding":false,"raw_author_name":"Chakraborty, Tanmoy","raw_affiliation_strings":["IIT Delhi"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"IIT Delhi","institution_ids":["https://openalex.org/I68891433"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":["https://openalex.org/I68891433"],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":true,"cited_by_count":0,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":null,"last_page":null},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10028","display_name":"Topic Modeling","score":0.7221999764442444,"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":0.7221999764442444,"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":0.23489999771118164,"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":0.013799999840557575,"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/test-time-scaling","display_name":"test-time scaling","score":0.9959999918937683},{"id":"https://openalex.org/keywords/multilingual-reasoning","display_name":"multilingual reasoning","score":0.9079999923706055},{"id":"https://openalex.org/keywords/prefix-tuning","display_name":"prefix tuning","score":0.8100000023841858},{"id":"https://openalex.org/keywords/low-resource-languages","display_name":"low-resource languages","score":0.6000000238418579},{"id":"https://openalex.org/keywords/internal-consistency","display_name":"internal consistency","score":0.3959999978542328},{"id":"https://openalex.org/keywords/deepseek-r1","display_name":"DeepSeek-R1","score":0.5799999833106995},{"id":"https://openalex.org/keywords/qwen","display_name":"Qwen","score":0.6499999761581421},{"id":"https://openalex.org/keywords/llm-reasoning","display_name":"LLM reasoning","score":0.3889999985694885}],"concepts":[{"id":"https://openalex.org/C99844830","wikidata":"https://www.wikidata.org/wiki/Q102441924","display_name":"Scaling","level":2,"score":0.7210999727249146},{"id":"https://openalex.org/C46686674","wikidata":"https://www.wikidata.org/wiki/Q466303","display_name":"Boosting (machine learning)","level":2,"score":0.6093000173568726},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5954999923706055},{"id":"https://openalex.org/C141603448","wikidata":"https://www.wikidata.org/wiki/Q134830","display_name":"Prefix","level":2,"score":0.5370000004768372},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5223000049591064},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.5223000049591064},{"id":"https://openalex.org/C2776436953","wikidata":"https://www.wikidata.org/wiki/Q5163215","display_name":"Consistency (knowledge bases)","level":2,"score":0.519599974155426},{"id":"https://openalex.org/C41895202","wikidata":"https://www.wikidata.org/wiki/Q8162","display_name":"Linguistics","level":1,"score":0.36890000104904175},{"id":"https://openalex.org/C180747234","wikidata":"https://www.wikidata.org/wiki/Q23373","display_name":"Cognitive psychology","level":1,"score":0.2833999991416931},{"id":"https://openalex.org/C150899416","wikidata":"https://www.wikidata.org/wiki/Q1820378","display_name":"Transfer of learning","level":2,"score":0.2800000011920929},{"id":"https://openalex.org/C2777379011","wikidata":"https://www.wikidata.org/wiki/Q938545","display_name":"Implicit learning","level":3,"score":0.26739999651908875},{"id":"https://openalex.org/C20162079","wikidata":"https://www.wikidata.org/wiki/Q1151406","display_name":"Case-based reasoning","level":2,"score":0.26660001277923584},{"id":"https://openalex.org/C2985612853","wikidata":"https://www.wikidata.org/wiki/Q185816","display_name":"Analogical reasoning","level":3,"score":0.26600000262260437}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2505.15508","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2505.15508","pdf_url":"https://arxiv.org/pdf/2505.15508","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.2505.15508","doi":"https://doi.org/10.48550/arxiv.2505.15508","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2505.15508","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":null,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2505.15508","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2505.15508","pdf_url":"https://arxiv.org/pdf/2505.15508","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"},"sustainable_development_goals":[],"sustainable_development_goals_aurora":[],"study_designs":[],"awards":[],"funders":[],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4415329122.pdf","grobid_xml":"https://content.openalex.org/works/W4415329122.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Test-time":[0],"scaling":[1,40,64,138],"has":[2,17],"emerged":[3],"as":[4],"a":[5],"widely":[6],"adopted":[7],"inference-time":[8],"strategy":[9],"for":[10,155],"boosting":[11],"reasoning":[12,93,127,133,146,152],"performance.":[13,147],"However,":[14],"its":[15,25],"effectiveness":[16],"been":[18],"studied":[19],"almost":[20],"exclusively":[21],"in":[22,27,41,108,144],"English,":[23],"leaving":[24],"behavior":[26],"other":[28],"languages":[29,88],"largely":[30],"unexplored.":[31],"We":[32,83],"present":[33],"the":[34,59],"first":[35],"systematic":[36],"study":[37],"of":[38],"test-time":[39,63,137],"multilingual":[42,145],"settings,":[43],"evaluating":[44],"DeepSeek-R1-Distill-LLama-8B":[45],"and":[46,51,125],"DeepSeek-R1-Distill-Qwen-7B":[47],"across":[48,67,106,139],"both":[49],"high-":[50],"low-resource":[52,87],"Latin-script":[53],"languages.":[54,68,157],"Our":[55],"findings":[56],"reveal":[57],"that":[58,86,95,130],"relative":[60],"gains":[61],"from":[62,98],"vary":[65],"significantly":[66,97,149],"Additionally,":[69],"models":[70],"frequently":[71],"switch":[72],"to":[73,135],"English":[74,99],"mid-reasoning,":[75],"even":[76],"when":[77],"operating":[78],"under":[79],"strictly":[80],"monolingual":[81],"prompts.":[82],"further":[84],"show":[85],"not":[89],"only":[90],"produce":[91],"initial":[92],"thoughts":[94],"differ":[96],"but":[100],"also":[101],"have":[102],"lower":[103],"internal":[104],"consistency":[105],"generations":[107],"their":[109],"early":[110],"reasoning.":[111],"Building":[112],"on":[113],"our":[114],"findings,":[115],"we":[116],"introduce":[117],"MITT":[118,148],"(Multilingual":[119],"Initial":[120],"Thought":[121],"Transfer),":[122],"an":[123],"unsupervised":[124],"lightweight":[126],"prefix-tuning":[128],"approach":[129],"transfers":[131],"high-resource":[132],"prefixes":[134],"enhance":[136],"all":[140],"languages,":[141],"addressing":[142],"inconsistencies":[143],"boosts":[150],"DeepSeek-R1-Distill-Qwen-7B's":[151],"performance,":[153],"especially":[154],"underrepresented":[156]},"counts_by_year":[],"updated_date":"2026-09-30T09:27:27.532063","created_date":"2025-10-19T00:00:00"}