{"id":"W3196239222","doi":"10.1145/3468264.3468591","title":"A comprehensive study of deep learning compiler bugs","year":2021,"lang":"en","type":"article","venue":"","topic":"Software Testing and Debugging Techniques","field":"Computer Science","cited_by":112,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Waterloo","funders":"National Key Research and Development Program of China; National Natural Science Foundation of China","keywords":"Compiler; Computer science; Programming language; Boosting (machine learning); Deep learning; Context (archaeology); Code (set theory); Code generation; Parallel computing; Artificial intelligence; Operating system","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":[0.00007800895,0.0000830554,0.000171662,0.00005736224,0.00007637463,0.0000512642,0.0003457524,0.00002566531,0.00002372636],"category_scores_gemma":[0.0001399718,0.00007575578,0.0000345037,0.0003802598,0.00001956192,0.00009115512,0.0003850565,0.0001272768,0.00001196421],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.000009669976,"about_ca_system_score_gemma":0.00002264069,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0001015481,"about_ca_topic_score_gemma":0.000009364899,"domain_scores_codex":[0.9991022,0.0001231756,0.0001773851,0.0002602906,0.0002027551,0.0001341524],"domain_scores_gemma":[0.9988021,0.0003762879,0.00006418298,0.0004276053,0.0002909381,0.0000389421],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.00002062044,0.004323571,0.4403572,0.0001577613,0.0002639222,0.0007527735,0.02967495,0.003232036,0.005186387,0.01674093,0.01756378,0.4817261],"study_design_scores_gemma":[0.003602179,0.003839559,0.3778014,0.0002166713,0.00006194551,0.0004472567,0.004212062,0.5353408,0.01995663,0.04398524,0.008983401,0.001552788],"study_design_candidate":"observational","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.330675,0.0001411585,0.6591977,0.00009361526,0.0001102752,0.00011171,1.096487e-7,0.006527221,0.003143224],"genre_scores_gemma":[0.8362119,0.000001869784,0.1634878,0.0001396571,0.00001024636,0.00000532713,6.100282e-7,0.000004834847,0.0001376912],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.5321088,"threshold_uncertainty_score":0.3089231,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.03173090442745961,"score_gpt":0.283983725789503,"score_spread":0.2522528213620434,"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."}}