{"id":"W3172592804","doi":"10.3929/ethz-b-000489720","title":"Formally Validating a Practical Verification Condition Generator (extended version)","year":2021,"lang":"en","type":"preprint","venue":"arXiv (Cornell University)","topic":"Security and Verification in Computing","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of British Columbia","funders":"","keywords":"Correctness; Soundness; Computer science; Programming language; Generator (circuit theory); Certificate; Implementation; Transformation (genetics); Control flow; Debugging; Theoretical computer science","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":["metaepi_narrow"],"consensus_categories":[],"category_scores_codex":[0.0004952087,0.0002794741,0.0002772298,0.000195967,0.0004323556,0.00052847,0.001096815,0.0003544971,0.00009381349],"category_scores_gemma":[0.0001987311,0.0003680984,0.0002034238,0.0006433725,0.00007491846,0.001205001,0.00139066,0.0007244422,0.0001222485],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0003096294,"about_ca_system_score_gemma":0.0006292175,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00003800847,"about_ca_topic_score_gemma":0.000006260615,"domain_scores_codex":[0.9975788,0.0003341276,0.0003309091,0.001230151,0.0001851023,0.0003408893],"domain_scores_gemma":[0.9973136,0.0002130366,0.000498125,0.001312385,0.0004843728,0.000178441],"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":[0.00004894333,0.0004622813,0.0006389807,0.0002411905,0.0002123269,0.0005145976,0.001548018,0.07832704,0.001903561,0.912336,0.001704962,0.002062115],"study_design_scores_gemma":[0.0004803918,0.00004336301,0.001182116,0.0001195758,0.00006912054,0.00003268711,0.0004194882,0.9848179,0.003545655,0.007320765,0.001438774,0.0005301143],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.2688062,0.00003020295,0.7284834,0.0002898876,0.00090454,0.0001940516,0.00000890434,0.0002814964,0.001001252],"genre_scores_gemma":[0.9659409,0.00007615459,0.033205,0.0002439519,0.000168265,0.00000240477,0.0002078183,0.00001545653,0.0001400421],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.9064909,"threshold_uncertainty_score":0.9998771,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.1001160946931149,"score_gpt":0.2374576539762914,"score_spread":0.1373415592831765,"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."}}