{"id":"W4409362569","doi":"10.1609/aaai.v39i24.34717","title":"CodeHalu: Investigating Code Hallucinations in LLMs via Execution-based Verification","year":2025,"lang":"en","type":"article","venue":"Proceedings of the AAAI Conference on Artificial Intelligence","topic":"Security and Verification in Computing","field":"Computer Science","cited_by":13,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Alberta; Mila - Quebec Artificial Intelligence Institute","funders":"","keywords":"Code (set theory); Computer science; Programming language; Psychology","routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":false},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008103742,0.0001723806,0.0001967215,0.0003073757,0.0003013822,0.0002363739,0.001774056,0.0001041923,0.000015548],"category_scores_gemma":[0.0008604208,0.0001589008,0.00007551725,0.001765295,0.0002518792,0.0003477796,0.0001870627,0.0003314321,0.00002596947],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001191791,"about_ca_system_score_gemma":0.0002617684,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0000875726,"about_ca_topic_score_gemma":0.00003690324,"domain_scores_codex":[0.998147,0.00003971496,0.0006886079,0.00050646,0.0003589953,0.0002591856],"domain_scores_gemma":[0.9982631,0.0002397629,0.0003891315,0.0003746627,0.00068173,0.00005165415],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.000008787762,0.0001323966,0.0009152325,0.00004288988,0.000004521387,8.76229e-8,0.0008855358,0.001337901,0.02594906,0.9462171,0.000062035,0.0244445],"study_design_scores_gemma":[0.00002992226,0.00002722753,0.001535426,0.0002926883,0.000004018615,4.479178e-7,0.0002147768,0.5737259,0.26322,0.1607694,0.000068441,0.000111821],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.1420108,0.00003728789,0.8241848,0.01826999,0.0007707653,0.0007085774,0.000004953498,0.0002055073,0.01380737],"genre_scores_gemma":[0.9886951,0.000006682264,0.0106259,0.000519666,0.00002269432,0.00005185614,0.000001324235,0.000005911197,0.0000708645],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.8466843,"threshold_uncertainty_score":0.6479787,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.07942367299217998,"score_gpt":0.320831477344202,"score_spread":0.241407804352022,"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."}}