{"id":"https://openalex.org/W7148516234","doi":"https://doi.org/10.48550/arxiv.2604.00438","title":"TR-ICRL: Test-Time Rethinking for In-Context Reinforcement Learning","display_name":"TR-ICRL: Test-Time Rethinking for In-Context Reinforcement Learning","publication_year":2026,"publication_date":"2026-04-01","ids":{"openalex":"https://openalex.org/W7148516234","doi":"https://doi.org/10.48550/arxiv.2604.00438"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2604.00438","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2604.00438","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"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":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"article"},"type":"preprint","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.48550/arxiv.2604.00438","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5132807617","display_name":"Wenxuan Jiang","orcid":null},"institutions":[],"countries":[],"is_corresponding":true,"raw_author_name":"Jiang, Wenxuan","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5132819158","display_name":"Yuxin Zuo","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zuo, Yuxin","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5132826799","display_name":"Zijian Zhang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhang, Zijian","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5132826873","display_name":"Xuecheng Wu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wu, Xuecheng","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5132759104","display_name":"Zining Fan","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Fan, Zining","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5132795873","display_name":"Wenxuan Liu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Liu, Wenxuan","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5132792114","display_name":"Li Chen","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Chen, Li","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5132805734","display_name":"Xiaoyu Li","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Li, Xiaoyu","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5132791146","display_name":"Xuezhi Cao","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Cao, Xuezhi","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5132814972","display_name":"Xiaolong Jin","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jin, Xiaolong","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5132788974","display_name":"Ninghao Liu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Liu, Ninghao","raw_affiliation_strings":[],"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":11,"corresponding_author_ids":["https://openalex.org/A5132807617"],"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/T11714","display_name":"Multimodal Machine Learning Applications","score":0.3598000109195709,"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"}},"topics":[{"id":"https://openalex.org/T11714","display_name":"Multimodal Machine Learning Applications","score":0.3598000109195709,"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"}},{"id":"https://openalex.org/T13702","display_name":"Machine Learning in Healthcare","score":0.12269999831914902,"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":0.11270000040531158,"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/reinforcement-learning","display_name":"Reinforcement learning","score":0.7361999750137329},{"id":"https://openalex.org/keywords/set","display_name":"Set (abstract data type)","score":0.5083000063896179},{"id":"https://openalex.org/keywords/robustness","display_name":"Robustness (evolution)","score":0.4636000096797943},{"id":"https://openalex.org/keywords/formative-assessment","display_name":"Formative assessment","score":0.3734000027179718},{"id":"https://openalex.org/keywords/context","display_name":"Context (archaeology)","score":0.36649999022483826},{"id":"https://openalex.org/keywords/language-model","display_name":"Language model","score":0.34940001368522644}],"concepts":[{"id":"https://openalex.org/C97541855","wikidata":"https://www.wikidata.org/wiki/Q830687","display_name":"Reinforcement learning","level":2,"score":0.7361999750137329},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7078999876976013},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.6031000018119812},{"id":"https://openalex.org/C177264268","wikidata":"https://www.wikidata.org/wiki/Q1514741","display_name":"Set (abstract data type)","level":2,"score":0.5083000063896179},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.47909998893737793},{"id":"https://openalex.org/C63479239","wikidata":"https://www.wikidata.org/wiki/Q7353546","display_name":"Robustness (evolution)","level":3,"score":0.4636000096797943},{"id":"https://openalex.org/C42525527","wikidata":"https://www.wikidata.org/wiki/Q1209955","display_name":"Formative assessment","level":2,"score":0.3734000027179718},{"id":"https://openalex.org/C2779343474","wikidata":"https://www.wikidata.org/wiki/Q3109175","display_name":"Context (archaeology)","level":2,"score":0.36649999022483826},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.34940001368522644},{"id":"https://openalex.org/C2776760102","wikidata":"https://www.wikidata.org/wiki/Q5139990","display_name":"Code (set theory)","level":3,"score":0.32499998807907104},{"id":"https://openalex.org/C43126263","wikidata":"https://www.wikidata.org/wiki/Q128751","display_name":"Source code","level":2,"score":0.31189998984336853},{"id":"https://openalex.org/C2780148112","wikidata":"https://www.wikidata.org/wiki/Q1432581","display_name":"Proxy (statistics)","level":2,"score":0.29809999465942383},{"id":"https://openalex.org/C67203356","wikidata":"https://www.wikidata.org/wiki/Q1321905","display_name":"Reinforcement","level":2,"score":0.28110000491142273},{"id":"https://openalex.org/C177877439","wikidata":"https://www.wikidata.org/wiki/Q7604413","display_name":"Statistical relational learning","level":3,"score":0.2806999981403351},{"id":"https://openalex.org/C46686674","wikidata":"https://www.wikidata.org/wiki/Q466303","display_name":"Boosting (machine learning)","level":2,"score":0.2606000006198883},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.2551000118255615}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2604.00438","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2604.00438","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"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":null,"is_accepted":false,"is_published":null,"raw_source_name":null,"raw_type":"article"}],"best_oa_location":{"id":"doi:10.48550/arxiv.2604.00438","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2604.00438","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"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":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"article"},"sustainable_development_goals":[{"id":"https://metadata.un.org/sdg/4","display_name":"Quality Education","score":0.4859820306301117}],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"In-Context":[0,47],"Reinforcement":[1,48],"Learning":[2,49],"(ICRL)":[3],"enables":[4],"Large":[5],"Language":[6],"Models":[7],"(LLMs)":[8],"to":[9,34,113,139],"learn":[10],"online":[11],"from":[12,71,100],"external":[13],"rewards":[14],"directly":[15],"within":[16],"the":[17,67,127,136,145,192],"context":[18],"window.":[19],"However,":[20],"a":[21,51,77,86,96,111,141,149],"central":[22],"challenge":[23],"in":[24],"ICRL":[25,53,82],"is":[26,98,133,156,201],"reward":[27,115],"estimation,":[28],"as":[29,110],"models":[30],"typically":[31],"lack":[32],"access":[33],"ground-truths":[35],"during":[36],"inference.":[37],"To":[38],"address":[39],"this":[40,101,129],"limitation,":[41],"we":[42],"propose":[43],"Test-Time":[44],"Rethinking":[45],"for":[46,56,76,91],"(TR-ICRL),":[50],"novel":[52],"framework":[54],"designed":[55],"both":[57],"reasoning":[58,160],"and":[59,117,161,180,188,194],"knowledge-intensive":[60,162],"tasks.":[61],"TR-ICRL":[62,155,171],"operates":[63],"by":[64,174],"first":[65],"retrieving":[66],"most":[68],"relevant":[69],"instances":[70],"an":[72],"unlabeled":[73],"evaluation":[74],"set":[75,87,102],"given":[78],"query.":[79],"During":[80],"each":[81],"iteration,":[83],"LLM":[84,122],"generates":[85],"of":[88,152,196],"candidate":[89],"answers":[90],"every":[92],"retrieved":[93],"instance.":[94],"Next,":[95],"pseudo-label":[97],"derived":[99],"through":[103,123,148],"majority":[104,153],"voting.":[105,154],"This":[106],"label":[107],"then":[108],"serves":[109],"proxy":[112],"give":[114],"messages":[116],"generate":[118],"formative":[119],"feedbacks,":[120],"guiding":[121],"iterative":[124],"refinement.":[125],"In":[126],"end,":[128],"synthesized":[130],"contextual":[131],"information":[132],"integrated":[134],"with":[135,144],"original":[137],"query":[138],"form":[140],"comprehensive":[142],"prompt,":[143],"answer":[146],"determining":[147],"final":[150],"round":[151],"evaluated":[157],"on":[158,176,178,183],"mainstream":[159],"tasks,":[163],"where":[164],"it":[165],"demonstrates":[166],"significant":[167],"performance":[168],"gains.":[169],"Remarkably,":[170],"improves":[172],"Qwen2.5-7B":[173],"21.23%":[175],"average":[177],"MedQA":[179],"even":[181],"137.59%":[182],"AIME2024.":[184],"Extensive":[185],"ablation":[186],"studies":[187],"analyses":[189],"further":[190],"validate":[191],"effectiveness":[193],"robustness":[195],"our":[197],"approach.":[198],"Our":[199],"code":[200],"available":[202],"at":[203],"https://github.com/pangpang-xuan/TR_ICRL.":[204]},"counts_by_year":[],"updated_date":"2026-04-03T16:44:17.987007","created_date":"2026-04-03T00:00:00"}