{"id":"https://openalex.org/W7134838181","doi":"https://doi.org/10.48550/arxiv.2603.08068","title":"In-Context Reinforcement Learning for Tool Use in Large Language Models","display_name":"In-Context Reinforcement Learning for Tool Use in Large Language Models","publication_year":2026,"publication_date":"2026-03-09","ids":{"openalex":"https://openalex.org/W7134838181","doi":"https://doi.org/10.48550/arxiv.2603.08068"},"language":null,"primary_location":{"id":"pmh:doi:10.48550/arxiv.2603.08068","is_oa":true,"landing_page_url":null,"pdf_url":null,"source":{"id":"https://openalex.org/S4406922384","display_name":"Open MIND","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"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":"Article"},"type":"preprint","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":null,"any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5128671339","display_name":"Yaoqi Ye","orcid":null},"institutions":[],"countries":[],"is_corresponding":true,"raw_author_name":"Ye, Yaoqi","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5128657300","display_name":"Yiran Zhao","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhao, Yiran","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5128689148","display_name":"Keyu Duan","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Duan, Keyu","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5128662338","display_name":"Zeyu Zheng","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zheng, Zeyu","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5128638223","display_name":"Kenji Kawaguchi","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Kawaguchi, Kenji","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5128657451","display_name":"Cihang Xie","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Xie, Cihang","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5128658372","display_name":"Michael Qizhe Shieh","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Shieh, Michael Qizhe","raw_affiliation_strings":[],"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":7,"corresponding_author_ids":["https://openalex.org/A5128671339"],"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/T10028","display_name":"Topic Modeling","score":0.38940000534057617,"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":0.38940000534057617,"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":0.16590000689029694,"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/T10181","display_name":"Natural Language Processing Techniques","score":0.06870000064373016,"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":0.7026000022888184},{"id":"https://openalex.org/keywords/language-model","display_name":"Language model","score":0.5318999886512756},{"id":"https://openalex.org/keywords/python","display_name":"Python (programming language)","score":0.4860000014305115},{"id":"https://openalex.org/keywords/language-understanding","display_name":"Language understanding","score":0.43540000915527344},{"id":"https://openalex.org/keywords/computation","display_name":"Computation","score":0.3598000109195709},{"id":"https://openalex.org/keywords/range","display_name":"Range (aeronautics)","score":0.3447999954223633}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7946000099182129},{"id":"https://openalex.org/C97541855","wikidata":"https://www.wikidata.org/wiki/Q830687","display_name":"Reinforcement learning","level":2,"score":0.7026000022888184},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.626800000667572},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.5318999886512756},{"id":"https://openalex.org/C519991488","wikidata":"https://www.wikidata.org/wiki/Q28865","display_name":"Python (programming language)","level":2,"score":0.4860000014305115},{"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/C2983448237","wikidata":"https://www.wikidata.org/wiki/Q1078276","display_name":"Language understanding","level":2,"score":0.43540000915527344},{"id":"https://openalex.org/C45374587","wikidata":"https://www.wikidata.org/wiki/Q12525525","display_name":"Computation","level":2,"score":0.3598000109195709},{"id":"https://openalex.org/C204323151","wikidata":"https://www.wikidata.org/wiki/Q905424","display_name":"Range (aeronautics)","level":2,"score":0.3447999954223633},{"id":"https://openalex.org/C122783720","wikidata":"https://www.wikidata.org/wiki/Q183065","display_name":"Interpreter","level":2,"score":0.30140000581741333},{"id":"https://openalex.org/C51632099","wikidata":"https://www.wikidata.org/wiki/Q3985153","display_name":"Training set","level":2,"score":0.29499998688697815},{"id":"https://openalex.org/C95623464","wikidata":"https://www.wikidata.org/wiki/Q1096149","display_name":"Classifier (UML)","level":2,"score":0.2766999900341034},{"id":"https://openalex.org/C136389625","wikidata":"https://www.wikidata.org/wiki/Q334384","display_name":"Supervised learning","level":3,"score":0.26930001378059387},{"id":"https://openalex.org/C107457646","wikidata":"https://www.wikidata.org/wiki/Q207434","display_name":"Human\u2013computer interaction","level":1,"score":0.25369998812675476}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:doi:10.48550/arxiv.2603.08068","is_oa":true,"landing_page_url":null,"pdf_url":null,"source":{"id":"https://openalex.org/S4406922384","display_name":"Open MIND","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"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":"Article"},{"id":"doi:10.48550/arxiv.2603.08068","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2603.08068","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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":null,"raw_source_name":null,"raw_type":"article"}],"best_oa_location":{"id":"pmh:doi:10.48550/arxiv.2603.08068","is_oa":true,"landing_page_url":null,"pdf_url":null,"source":{"id":"https://openalex.org/S4406922384","display_name":"Open MIND","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"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":"Article"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"While":[0],"large":[1],"language":[2],"models":[3,35,56],"(LLMs)":[4],"exhibit":[5],"strong":[6],"reasoning":[7,181],"abilities,":[8],"their":[9,21],"performance":[10],"on":[11,70],"complex":[12],"tasks":[13],"is":[14,31,96,157],"often":[15,86],"constrained":[16],"by":[17,80,120],"the":[18,116,125,136,141,152,166],"limitations":[19],"of":[20,90,128,154,180],"internal":[22],"knowledge.":[23],"A":[24],"compelling":[25],"approach":[26],"to":[27,32,57,98,139,144,169,200],"overcome":[28],"this":[29,103],"challenge":[30],"augment":[33],"these":[34,59],"with":[36,75],"external":[37,146],"tools":[38,60,171],"--":[39],"such":[40],"as":[41,149,195],"Python":[42],"interpreters":[43],"for":[44,50,93,118],"mathematical":[45],"computations":[46],"or":[47,100],"search":[48],"engines":[49],"retrieving":[51],"factual":[52],"information.":[53],"However,":[54],"enabling":[55],"use":[58],"effectively":[61],"remains":[62],"a":[63,162,178,196],"significant":[64],"challenge.":[65],"Existing":[66],"methods":[67],"typically":[68],"rely":[69],"cold-start":[71],"pipelines":[72],"that":[73,114,187],"begin":[74],"supervised":[76],"fine-tuning":[77],"(SFT),":[78],"followed":[79],"reinforcement":[81],"learning":[82],"(RL).":[83],"These":[84],"approaches":[85],"require":[87],"substantial":[88],"amounts":[89],"labeled":[91],"data":[92],"SFT,":[94],"which":[95],"expensive":[97],"annotate":[99],"synthesize.":[101],"In":[102],"work,":[104],"we":[105],"propose":[106],"In-Context":[107],"Reinforcement":[108],"Learning":[109],"(ICRL),":[110],"an":[111],"RL-only":[112],"framework":[113],"eliminates":[115],"need":[117],"SFT":[119],"leveraging":[121],"few-shot":[122],"prompting":[123],"during":[124],"rollout":[126,137],"stage":[127],"RL.":[129],"Specifically,":[130],"ICRL":[131,188],"introduces":[132],"in-context":[133,155],"examples":[134,156],"within":[135],"prompts":[138],"teach":[140],"model":[142,167],"how":[143],"invoke":[145],"tools.":[147],"Furthermore,":[148],"training":[150],"progresses,":[151],"number":[153],"gradually":[158],"reduced,":[159],"eventually":[160],"reaching":[161],"zero-shot":[163],"setting":[164],"where":[165],"learns":[168],"call":[170],"independently.":[172],"We":[173],"conduct":[174],"extensive":[175],"experiments":[176],"across":[177],"range":[179],"and":[182],"tool-use":[183],"benchmarks.":[184],"Results":[185],"show":[186],"achieves":[189],"state-of-the-art":[190],"performance,":[191],"demonstrating":[192],"its":[193],"effectiveness":[194],"scalable,":[197],"data-efficient":[198],"alternative":[199],"traditional":[201],"SFT-based":[202],"pipelines.":[203]},"counts_by_year":[],"updated_date":"2026-04-04T16:13:02.066488","created_date":"2026-03-11T00:00:00"}