{"id":"https://openalex.org/W7166738485","doi":"https://doi.org/10.48550/arxiv.2606.30044","title":"Building Multi-Task Agentic LLMs via Two-Phase Distillation","display_name":"Building Multi-Task Agentic LLMs via Two-Phase Distillation","publication_year":2026,"publication_date":"2026-06-29","ids":{"openalex":"https://openalex.org/W7166738485","doi":"https://doi.org/10.48550/arxiv.2606.30044"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2606.30044","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.30044","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,"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":"Preprint"},"type":"preprint","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.48550/arxiv.2606.30044","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5139675195","display_name":"Huaijie Wang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wang, Huaijie","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5053837667","display_name":"S Xu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Xu, Shusheng","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139643005","display_name":"Yi Wu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wu, Yi","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5139633075","display_name":"Kaifeng Lyu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Lyu, Kaifeng","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/T11714","display_name":"Multimodal Machine Learning Applications","score":0.18320000171661377,"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":0.18320000171661377,"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/T10462","display_name":"Reinforcement Learning in Robotics","score":0.1770000010728836,"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":0.17139999568462372,"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/distillation","display_name":"Distillation","score":0.791100025177002},{"id":"https://openalex.org/keywords/key","display_name":"Key (lock)","score":0.6700000166893005},{"id":"https://openalex.org/keywords/forcing","display_name":"Forcing (mathematics)","score":0.48919999599456787},{"id":"https://openalex.org/keywords/training","display_name":"Training (meteorology)","score":0.4228000044822693},{"id":"https://openalex.org/keywords/matching","display_name":"Matching (statistics)","score":0.39320001006126404}],"concepts":[{"id":"https://openalex.org/C204030448","wikidata":"https://www.wikidata.org/wiki/Q101017","display_name":"Distillation","level":2,"score":0.791100025177002},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.6700000166893005},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6238999962806702},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5031999945640564},{"id":"https://openalex.org/C197115733","wikidata":"https://www.wikidata.org/wiki/Q1003136","display_name":"Forcing (mathematics)","level":2,"score":0.48919999599456787},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.44119998812675476},{"id":"https://openalex.org/C2777211547","wikidata":"https://www.wikidata.org/wiki/Q17141490","display_name":"Training (meteorology)","level":2,"score":0.4228000044822693},{"id":"https://openalex.org/C165064840","wikidata":"https://www.wikidata.org/wiki/Q1321061","display_name":"Matching (statistics)","level":2,"score":0.39320001006126404},{"id":"https://openalex.org/C51632099","wikidata":"https://www.wikidata.org/wiki/Q3985153","display_name":"Training set","level":2,"score":0.32409998774528503},{"id":"https://openalex.org/C154030694","wikidata":"https://www.wikidata.org/wiki/Q1436074","display_name":"Fractionating column","level":3,"score":0.29109999537467957}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2606.30044","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.30044","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,"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":"Preprint"}],"best_oa_location":{"id":"doi:10.48550/arxiv.2606.30044","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.30044","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,"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":"Preprint"},"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":{"A":[0],"key":[1],"step":[2],"toward":[3],"artificial":[4],"general":[5],"intelligence":[6],"is":[7,103],"to":[8,22,44,63,90,96,152],"train":[9],"models":[10,25],"that":[11,55,82,132],"can":[12,83],"perform":[13],"multiple":[14,73],"tasks.":[15,52],"In":[16,99],"this":[17,133,154],"paper,":[18],"we":[19,113],"study":[20],"how":[21],"build":[23],"such":[24],"by":[26,110,121],"first":[27],"training":[28,46],"separate":[29],"RL":[30,138],"experts":[31],"for":[32,141],"individual":[33,143],"tasks":[34,74],"and":[35,94,128],"then":[36],"consolidating":[37],"them":[38],"via":[39],"distillation,":[40],"as":[41],"an":[42],"alternative":[43],"directly":[45],"a":[47,76,115],"single":[48],"model":[49],"on":[50],"mixed":[51],"We":[53],"show":[54],"off-policy":[56,118,146],"distillation":[57,102,119,149],"degrades":[58],"in":[59],"multi-task":[60],"settings":[61],"due":[62],"the":[64,85],"mode-covering":[65],"nature":[66],"of":[67,79],"forward":[68],"KL:":[69],"aggregating":[70],"data":[71],"from":[72],"introduces":[75],"large":[77],"number":[78],"behavioral":[80],"modes":[81],"exceed":[84],"student's":[86],"capacity,":[87],"forcing":[88],"it":[89],"average":[91],"across":[92,125],"behaviors":[93],"leading":[95],"degraded":[97],"performance.":[98,155],"contrast,":[100],"on-policy":[101,122,148],"mode-seeking":[104],"but":[105],"requires":[106],"strong":[107],"initialization.":[108],"Inspired":[109],"these":[111],"observations,":[112],"propose":[114],"two-phase":[116,134],"approach:":[117],"followed":[120],"refinement.":[123],"Evaluation":[124],"conversational":[126],"agents":[127],"text-based":[129],"games":[130],"confirms":[131],"approach":[135],"matches":[136],"single-task":[137],"expert":[139],"performance":[140],"each":[142],"task,":[144],"whereas":[145],"or":[147],"alone":[150],"fails":[151],"match":[153]},"counts_by_year":[],"updated_date":"2026-07-28T07:46:37.118299","created_date":"2026-07-01T00:00:00"}