{"id":"https://openalex.org/W7162518462","doi":"https://doi.org/10.48550/arxiv.2605.26720","title":"Towards Feedback-to-Plan Decisions for Self-Evolving LLM Agents in CUDA Kernel Generation","display_name":"Towards Feedback-to-Plan Decisions for Self-Evolving LLM Agents in CUDA Kernel Generation","publication_year":2026,"publication_date":"2026-05-26","ids":{"openalex":"https://openalex.org/W7162518462","doi":"https://doi.org/10.48550/arxiv.2605.26720"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2605.26720","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.26720","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":"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"},"type":"preprint","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.48550/arxiv.2605.26720","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5137113940","display_name":"Yee Hin Chong","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Chong, Yee Hin","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5137161940","display_name":"Jiaming Wu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wu, Jiaming","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5137156249","display_name":"Youhui Zhang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhang, Youhui","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5137117943","display_name":"Peng Qu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Qu, Peng","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/T10462","display_name":"Reinforcement Learning in Robotics","score":0.2685000002384186,"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/T10462","display_name":"Reinforcement Learning in Robotics","score":0.2685000002384186,"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/T10906","display_name":"AI-based Problem Solving and Planning","score":0.19580000638961792,"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":0.1624000072479248,"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"}}],"keywords":[{"id":"https://openalex.org/keywords/kernel","display_name":"Kernel (algebra)","score":0.49970000982284546},{"id":"https://openalex.org/keywords/trajectory","display_name":"Trajectory","score":0.49799999594688416},{"id":"https://openalex.org/keywords/automated-planning-and-scheduling","display_name":"Automated planning and scheduling","score":0.3930000066757202},{"id":"https://openalex.org/keywords/attribution","display_name":"Attribution","score":0.35929998755455017},{"id":"https://openalex.org/keywords/cuda","display_name":"CUDA","score":0.3239000141620636},{"id":"https://openalex.org/keywords/local-planning","display_name":"Local planning","score":0.29170000553131104}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7210000157356262},{"id":"https://openalex.org/C74193536","wikidata":"https://www.wikidata.org/wiki/Q574844","display_name":"Kernel (algebra)","level":2,"score":0.49970000982284546},{"id":"https://openalex.org/C13662910","wikidata":"https://www.wikidata.org/wiki/Q193139","display_name":"Trajectory","level":2,"score":0.49799999594688416},{"id":"https://openalex.org/C114073186","wikidata":"https://www.wikidata.org/wiki/Q2631895","display_name":"Automated planning and scheduling","level":2,"score":0.3930000066757202},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.3711000084877014},{"id":"https://openalex.org/C143299363","wikidata":"https://www.wikidata.org/wiki/Q900584","display_name":"Attribution","level":2,"score":0.35929998755455017},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.3573000133037567},{"id":"https://openalex.org/C2778119891","wikidata":"https://www.wikidata.org/wiki/Q477690","display_name":"CUDA","level":2,"score":0.3239000141620636},{"id":"https://openalex.org/C2988400853","wikidata":"https://www.wikidata.org/wiki/Q69883","display_name":"Local planning","level":2,"score":0.29170000553131104},{"id":"https://openalex.org/C120936955","wikidata":"https://www.wikidata.org/wiki/Q2155640","display_name":"Empirical research","level":2,"score":0.29120001196861267},{"id":"https://openalex.org/C63479239","wikidata":"https://www.wikidata.org/wiki/Q7353546","display_name":"Robustness (evolution)","level":3,"score":0.28529998660087585},{"id":"https://openalex.org/C77618280","wikidata":"https://www.wikidata.org/wiki/Q1155772","display_name":"Scheme (mathematics)","level":2,"score":0.2775000035762787},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.2736000120639801},{"id":"https://openalex.org/C2776175482","wikidata":"https://www.wikidata.org/wiki/Q1195816","display_name":"Transfer (computing)","level":2,"score":0.2581999897956848},{"id":"https://openalex.org/C3017962763","wikidata":"https://www.wikidata.org/wiki/Q427492","display_name":"Feedback regulation","level":2,"score":0.2540000081062317}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2605.26720","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.26720","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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":null,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"doi:10.48550/arxiv.2605.26720","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.26720","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":"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"},"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":{"Large":[0],"language":[1],"models":[2,121],"(LLMs)":[3],"have":[4],"shown":[5],"strong":[6],"empirical":[7],"gains":[8],"as":[9,42],"self-evolving":[10],"agents":[11],"for":[12,62],"CUDA":[13],"kernel":[14],"generation,":[15],"driven":[16],"by":[17],"feedback-conditioned":[18],"planning":[19,24,44,67,98,108],"across":[20,131],"generations.":[21],"However,":[22],"how":[23],"decisions":[25,68],"attribute":[26],"and":[27,48,75,84,91,114,136],"combine":[28],"heterogeneous":[29],"feedback":[30,50,70,77,89,103],"signals":[31],"remains":[32],"opaque.":[33],"Standard":[34],"end-to-end":[35],"ablations":[36],"fail":[37],"to":[38,69,125],"resolve":[39],"this":[40],"question,":[41],"iterative":[43],"amplifies":[45],"early":[46],"perturbations":[47],"conflates":[49],"effects":[51,90],"with":[52],"trajectory-dependent":[53],"drift.":[54],"We":[55],"introduce":[56],"\\texttt{CUDAnalyst},":[57],"a":[58],"unified":[59],"analysis":[60],"layer":[61],"controlled,":[63],"generation-level":[64,82],"attribution":[65,87],"of":[66,88],"components":[71],"via":[72],"trajectory":[73],"freezing":[74],"selective":[76],"injection.":[78],"\\texttt{CUDAnalyst}":[79],"enables":[80],"stable":[81],"evaluation":[83],"principled":[85],"coalitional-style":[86],"interactions.":[92],"Our":[93],"results":[94],"show":[95],"that":[96,106,115,141],"explicit":[97],"is":[99,104,146],"beneficial":[100],"only":[101],"when":[102],"aligned,":[105],"effective":[107],"emerges":[109],"from":[110,118],"structured":[111],"multi-feedback":[112],"interactions,":[113],"high-level":[116],"plans":[117],"stronger":[119],"reasoning":[120],"can":[122],"partially":[123],"transfer":[124],"weaker":[126],"ones.":[127],"These":[128],"trends":[129],"hold":[130],"reference":[132,137],"backbones,":[133],"representative":[134],"workloads,":[135],"induction":[138],"regimes,":[139],"indicating":[140],"the":[142,149],"identified":[143],"feedback-to-plan":[144],"structure":[145],"robust":[147],"within":[148],"controlled":[150],"axes":[151],"studied.":[152]},"counts_by_year":[],"updated_date":"2026-07-28T07:46:37.118299","created_date":"2026-05-28T00:00:00"}