{"id":"https://openalex.org/W4417095965","doi":"https://doi.org/10.48550/arxiv.2506.05574","title":"When can in-context learning generalize out of task distribution?","display_name":"When can in-context learning generalize out of task distribution?","publication_year":2025,"publication_date":"2025-06-05","ids":{"openalex":"https://openalex.org/W4417095965","doi":"https://doi.org/10.48550/arxiv.2506.05574"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2506.05574","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2506.05574","pdf_url":"https://arxiv.org/pdf/2506.05574","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":null,"license_id":null,"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/2506.05574","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5113641753","display_name":"Chase Goddard","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Goddard, Chase","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5075482591","display_name":"Lindsay M. Smith","orcid":"https://orcid.org/0000-0001-5379-7699","observed_orcids":["https://orcid.org/0000-0001-5379-7699"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Smith, Lindsay M.","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5081526793","display_name":"Vudtiwat Ngampruetikorn","orcid":"https://orcid.org/0000-0001-8548-3029","observed_orcids":["https://orcid.org/0000-0001-8548-3029"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ngampruetikorn, Vudtiwat","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5060296717","display_name":"David J. Schwab","orcid":"https://orcid.org/0000-0003-2726-4527","observed_orcids":["https://orcid.org/0000-0003-2726-4527"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Schwab, David J.","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":0,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":false,"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/T11307","display_name":"Domain Adaptation and Few-Shot Learning","score":0.9824000000953674,"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/T11307","display_name":"Domain Adaptation and Few-Shot Learning","score":0.9824000000953674,"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":0.009499999694526196,"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/T12072","display_name":"Machine Learning and Algorithms","score":0.00139999995008111,"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/in-context-learning","display_name":"in-context learning","score":1.0},{"id":"https://openalex.org/keywords/out-of-distribution-generalization","display_name":"out-of-distribution generalization","score":0.9660000205039978},{"id":"https://openalex.org/keywords/task-variety","display_name":"task variety","score":0.8579999804496765},{"id":"https://openalex.org/keywords/pre-trained-transformers","display_name":"pre-trained transformer models","score":0.7730000019073486},{"id":"https://openalex.org/keywords/linear-regression","display_name":"linear regression","score":0.6660000085830688},{"id":"https://openalex.org/keywords/phase-transitions","display_name":"phase transitions","score":0.6010000109672546},{"id":"https://openalex.org/keywords/model-specialization","display_name":"model specialization","score":0.4659999907016754},{"id":"https://openalex.org/keywords/phase-diagram","display_name":"phase diagram","score":0.5630000233650208},{"id":"https://openalex.org/keywords/model-depth","display_name":"model depth","score":0.597000002861023}],"concepts":[{"id":"https://openalex.org/C111030470","wikidata":"https://www.wikidata.org/wiki/Q1430460","display_name":"Curse of dimensionality","level":2,"score":0.6399999856948853},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.57669997215271},{"id":"https://openalex.org/C66322947","wikidata":"https://www.wikidata.org/wiki/Q11658","display_name":"Transformer","level":3,"score":0.548799991607666},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5382999777793884},{"id":"https://openalex.org/C2780451532","wikidata":"https://www.wikidata.org/wiki/Q759676","display_name":"Task (project management)","level":2,"score":0.5266000032424927},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.4472000002861023},{"id":"https://openalex.org/C175154964","wikidata":"https://www.wikidata.org/wiki/Q380077","display_name":"Task analysis","level":3,"score":0.38659998774528503},{"id":"https://openalex.org/C28006648","wikidata":"https://www.wikidata.org/wiki/Q6934509","display_name":"Multi-task learning","level":3,"score":0.3693999946117401},{"id":"https://openalex.org/C2780801425","wikidata":"https://www.wikidata.org/wiki/Q5164392","display_name":"Construct (python library)","level":2,"score":0.3296999931335449},{"id":"https://openalex.org/C83546350","wikidata":"https://www.wikidata.org/wiki/Q1139051","display_name":"Regression","level":2,"score":0.29750001430511475},{"id":"https://openalex.org/C158622935","wikidata":"https://www.wikidata.org/wiki/Q660848","display_name":"Nonlinear system","level":2,"score":0.29510000348091125},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.28130000829696655},{"id":"https://openalex.org/C70518039","wikidata":"https://www.wikidata.org/wiki/Q16000077","display_name":"Dimensionality reduction","level":2,"score":0.2784999907016754}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2506.05574","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2506.05574","pdf_url":"https://arxiv.org/pdf/2506.05574","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":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2506.05574","doi":"https://doi.org/10.48550/arxiv.2506.05574","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2506.05574","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":"pmh:oai:arXiv.org:2506.05574","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2506.05574","pdf_url":"https://arxiv.org/pdf/2506.05574","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":null,"license_id":null,"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":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W4415320283","https://openalex.org/W4403578025","https://openalex.org/W7133243359","https://openalex.org/W7163928116","https://openalex.org/W7084124427","https://openalex.org/W4388328818","https://openalex.org/W4414694875","https://openalex.org/W7214289539","https://openalex.org/W4411249741","https://openalex.org/W4415286575"],"abstract_inverted_index":{"In-context":[0],"learning":[1],"(ICL)":[2],"is":[3],"a":[4,21,59,86,89,102,142],"remarkable":[5],"capability":[6],"of":[7,48,62,69,107,119,129,150,157,170,176],"pretrained":[8],"transformers":[9,72,84],"that":[10,79],"allows":[11],"models":[12],"to":[13,15,36,65,101,109,145],"generalize":[14,39],"unseen":[16],"tasks":[17,50],"after":[18],"seeing":[19],"only":[20,95],"few":[22],"examples.":[23],"We":[24,77,114,140],"investigate":[25,116],"empirically":[26],"the":[27,31,46,53,67,97,110,117,120,124,130,155,168,171,174,177,181],"conditions":[28],"necessary":[29,51],"on":[30,45,74,126],"pretraining":[32,54,98,158],"distribution":[33,108],"for":[34],"ICL":[35,70,94],"emerge":[37],"and":[38,132,173],"\\emph{out-of-distribution}.":[40],"Previous":[41],"work":[42],"has":[43],"focused":[44],"number":[47,156],"distinct":[49],"in":[52,71,136],"dataset.":[55],"Here,":[56],"we":[57,162],"use":[58],"different":[60],"notion":[61],"task":[63,81,99,112,151],"diversity":[64,82,152],"study":[66],"emergence":[68],"trained":[73],"linear":[75],"functions.":[76],"find":[78],"as":[80,167],"increases,":[83],"undergo":[85],"transition":[87],"from":[88],"specialized":[90],"solution,":[91],"which":[92,104],"exhibits":[93],"within":[96],"distribution,":[100],"solution":[103],"generalizes":[105],"out":[106],"entire":[111],"space.":[113],"also":[115],"nature":[118],"solutions":[121],"learned":[122],"by":[123],"transformer":[125],"both":[127],"sides":[128],"transition,":[131],"observe":[133],"similar":[134],"transitions":[135],"nonlinear":[137],"regression":[138,178],"problems.":[139],"construct":[141],"phase":[143],"diagram":[144],"characterize":[146],"how":[147,164],"our":[148],"concept":[149],"interacts":[153],"with":[154],"tasks.":[159],"In":[160],"addition,":[161],"explore":[163],"factors":[165],"such":[166],"depth":[169],"model":[172],"dimensionality":[175],"problem":[179],"influence":[180],"transition.":[182]},"counts_by_year":[],"updated_date":"2026-08-26T07:47:46.906454","created_date":"2025-10-10T00:00:00"}