{"id":"W4399151229","doi":"10.48550/arxiv.2405.17503","title":"Code Repair with LLMs gives an Exploration-Exploitation Tradeoff","year":2024,"lang":"en","type":"preprint","venue":"arXiv (Cornell University)","topic":"Scientific Computing and Data Management","field":"Decision Sciences","cited_by":3,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"Natural Sciences and Engineering Research Council of Canada; Canadian Institute for Advanced Research; Cisco Systems; National Science Foundation","keywords":"Code (set theory); Computer science; Business; Computer security; Programming language","routes":{"ca_aff":false,"ca_fund":true,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":true},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.006691181,0.001174913,0.001054308,0.0009646689,0.0009694305,0.001986702,0.002164892,0.002098963,0.00484541],"category_scores_gemma":[0.03629677,0.0009589,0.001269117,0.0007536903,0.002636165,0.004046489,0.003144109,0.003405212,0.001576484],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001383348,"about_ca_system_score_gemma":0.00269661,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002369293,"about_ca_topic_score_gemma":0.005456469,"domain_scores_codex":[0.9951634,0.00200905,0.0002199642,0.0009035311,0.001294715,0.0004092218],"domain_scores_gemma":[0.9698398,0.02196509,0.001158253,0.00541292,0.001134948,0.0004889659],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0006584754,0.000431133,0.006727956,0.0005551885,0.0001469732,0.0003897017,0.001141279,0.5487937,0.01916143,0.1039317,0.008884613,0.3091778],"study_design_scores_gemma":[0.00007676119,0.0002024505,0.0003895613,0.00005920286,0.00004885915,0.000187199,0.0001565029,0.876985,0.005906123,0.1112658,0.00470079,0.00002174759],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.09019096,0.0005484591,0.8931796,0.003624821,0.00006242971,0.0001768731,0.000142394,0.004094472,0.007979935],"genre_scores_gemma":[0.4985833,0.0002083173,0.4937253,0.0007558111,0.00004579749,0.0002737993,0.0002751702,0.001155157,0.004977248],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.006691181,"threshold_uncertainty_score":0.0353868,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.3719729471813407,"score_gpt":0.2970075874439759,"score_spread":0.07496535973736479,"validation_status":"score_only:v0-immature-baseline","note":"Baseline scores from an immature model (maturity gate not passed). Scores rank; they never assert a category."}}