{"id":"https://openalex.org/W7166839634","doi":"https://doi.org/10.18653/v1/2026.findings-acl.462","title":"TT-SI: Self-Improving LLM Agents with Test-Time Training","display_name":"TT-SI: Self-Improving LLM Agents with Test-Time Training","publication_year":2026,"publication_date":"2026-01-01","ids":{"openalex":"https://openalex.org/W7166839634","doi":"https://doi.org/10.18653/v1/2026.findings-acl.462"},"language":null,"primary_location":{"id":"doi:10.18653/v1/2026.findings-acl.462","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-acl.462","pdf_url":"https://aclanthology.org/2026.findings-acl.462.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 2026","raw_type":"proceedings-article"},"type":"conference-paper","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://aclanthology.org/2026.findings-acl.462.pdf","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5038735910","display_name":"Emre Can Acikgoz","orcid":null},"institutions":[{"id":"https://openalex.org/I157725225","display_name":"University of Illinois Urbana-Champaign","ror":"https://ror.org/047426m28","country_code":"US","type":"education","lineage":["https://openalex.org/I157725225"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Emre Can Acikgoz","raw_affiliation_strings":["University of Illinois Urbana-Champaign"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of Illinois Urbana-Champaign","institution_ids":["https://openalex.org/I157725225"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139764426","display_name":"Cheng Qian","orcid":null},"institutions":[{"id":"https://openalex.org/I157725225","display_name":"University of Illinois Urbana-Champaign","ror":"https://ror.org/047426m28","country_code":"US","type":"education","lineage":["https://openalex.org/I157725225"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Cheng Qian","raw_affiliation_strings":["University of Illinois Urbana-Champaign"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of Illinois Urbana-Champaign","institution_ids":["https://openalex.org/I157725225"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139777803","display_name":"Heng Ji","orcid":null},"institutions":[{"id":"https://openalex.org/I157725225","display_name":"University of Illinois Urbana-Champaign","ror":"https://ror.org/047426m28","country_code":"US","type":"education","lineage":["https://openalex.org/I157725225"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Heng Ji","raw_affiliation_strings":["University of Illinois Urbana-Champaign"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of Illinois Urbana-Champaign","institution_ids":["https://openalex.org/I157725225"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139739701","display_name":"Dilek Hakkani-T\u00fcr","orcid":null},"institutions":[{"id":"https://openalex.org/I157725225","display_name":"University of Illinois Urbana-Champaign","ror":"https://ror.org/047426m28","country_code":"US","type":"education","lineage":["https://openalex.org/I157725225"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Dilek Hakkani-T\u00fcr","raw_affiliation_strings":["University of Illinois Urbana-Champaign"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of Illinois Urbana-Champaign","institution_ids":["https://openalex.org/I157725225"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5139768809","display_name":"Gokhan Tur","orcid":null},"institutions":[{"id":"https://openalex.org/I157725225","display_name":"University of Illinois Urbana-Champaign","ror":"https://ror.org/047426m28","country_code":"US","type":"education","lineage":["https://openalex.org/I157725225"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Gokhan Tur","raw_affiliation_strings":["University of Illinois Urbana-Champaign"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of Illinois Urbana-Champaign","institution_ids":["https://openalex.org/I157725225"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":["https://openalex.org/I157725225"],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":true,"cited_by_count":0,"citation_normalized_percentile":{"value":0.80515176,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"9483","last_page":"9508"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10906","display_name":"AI-based Problem Solving and Planning","score":0.07919999957084656,"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/T10906","display_name":"AI-based Problem Solving and Planning","score":0.07919999957084656,"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/T10456","display_name":"Multi-Agent Systems and Negotiation","score":0.07159999758005142,"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/T10462","display_name":"Reinforcement Learning in Robotics","score":0.046300001442432404,"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/training","display_name":"Training (meteorology)","score":0.43720000982284546},{"id":"https://openalex.org/keywords/action","display_name":"Action (physics)","score":0.2639000117778778},{"id":"https://openalex.org/keywords/work","display_name":"Work (physics)","score":0.2524999976158142},{"id":"https://openalex.org/keywords/perspective","display_name":"Perspective (graphical)","score":0.22859999537467957}],"concepts":[{"id":"https://openalex.org/C2777211547","wikidata":"https://www.wikidata.org/wiki/Q17141490","display_name":"Training (meteorology)","level":2,"score":0.43720000982284546},{"id":"https://openalex.org/C71924100","wikidata":"https://www.wikidata.org/wiki/Q11190","display_name":"Medicine","level":0,"score":0.41990000009536743},{"id":"https://openalex.org/C2780791683","wikidata":"https://www.wikidata.org/wiki/Q846785","display_name":"Action (physics)","level":2,"score":0.2639000117778778},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.26089999079704285},{"id":"https://openalex.org/C18762648","wikidata":"https://www.wikidata.org/wiki/Q42213","display_name":"Work (physics)","level":2,"score":0.2524999976158142},{"id":"https://openalex.org/C509550671","wikidata":"https://www.wikidata.org/wiki/Q126945","display_name":"Medical education","level":1,"score":0.2506999969482422},{"id":"https://openalex.org/C15744967","wikidata":"https://www.wikidata.org/wiki/Q9418","display_name":"Psychology","level":0,"score":0.24940000474452972},{"id":"https://openalex.org/C12713177","wikidata":"https://www.wikidata.org/wiki/Q1900281","display_name":"Perspective (graphical)","level":2,"score":0.22859999537467957},{"id":"https://openalex.org/C127413603","wikidata":"https://www.wikidata.org/wiki/Q11023","display_name":"Engineering","level":0,"score":0.20720000565052032},{"id":"https://openalex.org/C1862650","wikidata":"https://www.wikidata.org/wiki/Q186005","display_name":"Physical therapy","level":1,"score":0.20280000567436218}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.18653/v1/2026.findings-acl.462","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-acl.462","pdf_url":"https://aclanthology.org/2026.findings-acl.462.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 2026","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.18653/v1/2026.findings-acl.462","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-acl.462","pdf_url":"https://aclanthology.org/2026.findings-acl.462.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 2026","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[{"id":"https://openalex.org/G7559374616","display_name":null,"funder_award_id":"FA8650-23-C-7316","funder_id":"https://openalex.org/F4320332180","funder_display_name":"Defense Advanced Research Projects Agency"}],"funders":[{"id":"https://openalex.org/F4320332180","display_name":"Defense Advanced Research Projects Agency","ror":"https://ror.org/02caytj08"}],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7166839634.pdf","grobid_xml":"https://content.openalex.org/works/W7166839634.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"One":[0],"paradigm":[1,195],"of":[2,35,182,187],"language":[3],"model":[4,54,109],"(LM)":[5],"finetuning":[6],"relies":[7],"on":[8,41,165],"creating":[9],"large":[10,33],"training":[11,40,68,133],"datasets,":[12],"under":[13],"the":[14,52,106,158,180,185],"assumption":[15],"that":[16,51,108,155],"high":[17],"quantity":[18],"and":[19,39,91,124,170],"diversity":[20],"will":[21,55],"enable":[22],"models":[23],"to":[24,26,87],"generalize":[25,60],"novel":[27,71],"tasks":[28],"after":[29],"post-training.In":[30],"practice,":[31],"gathering":[32],"sets":[34],"data":[36,147],"is":[37,43,48],"inefficient,":[38],"them":[42],"prohibitively":[44],"expensive;":[45],"worse,":[46],"there":[47],"no":[49],"guarantee":[50],"resulting":[53,73],"handle":[56],"complex":[57],"scenarios":[58],"or":[59],"better.Moreover,":[61],"existing":[62],"techniques":[63],"rarely":[64],"assess":[65],"whether":[66],"a":[67,81,142,193],"sample":[69],"provides":[70],"information,":[72],"in":[74,99],"unnecessary":[75],"costs.In":[76],"this":[77],"work,":[78],"we":[79],"explore":[80,136],"new":[82,194],"Test-Time":[83,137],"Self-Improvement":[84],"(TT-SI)":[85],"algorithm":[86],"create":[88],"more":[89,176,198],"effective":[90],"generalizable":[92],"agentic":[93],"LMs":[94],"on-the-fly.TT-SI":[95],"can":[96],"be":[97],"summarized":[98],"three":[100],"steps:":[101],"(i)":[102],"first":[103],"it":[104],"identifies":[105],"samples":[107,121,130],"struggles":[110],"with":[111,160],"(self-awareness),":[112],"(ii)":[113],"then":[114],"generates":[115],"similar":[116],"examples":[117],"from":[118],"detected":[119],"uncertain":[120],"(self-data":[122],"augmentation),":[123],"(iii)":[125],"uses":[126],"these":[127],"newly":[128],"generated":[129],"at":[131,190],"testtime":[132],"(self-improvement).We":[134],"further":[135],"Distillation":[138],"(TT-D),":[139],"which":[140],"leverages":[141],"stronger":[143],"supervisor":[144],"for":[145,196],"targeted":[146],"generation.Empirical":[148],"evaluations":[149],"across":[150,167],"different":[151],"agent":[152],"benchmarks":[153,169],"demonstrate":[154],"TT-SI":[156],"improves":[157],"performance":[159],"+5.48%":[161],"absolute":[162],"accuracy":[163],"gain":[164],"average":[166],"all":[168],"surpasses":[171],"other":[172],"standard":[173],"learning":[174],"methods":[175],"efficiently.Our":[177],"findings":[178],"highlight":[179],"promise":[181],"TT-SI,":[183],"demonstrating":[184],"potential":[186],"self-improvement":[188],"algorithms":[189],"test-time":[191],"as":[192],"building":[197],"capable":[199],"agents":[200],"toward":[201],"self-evolution.":[202]},"counts_by_year":[],"updated_date":"2026-08-21T09:56:20.448147","created_date":"2026-07-02T00:00:00"}