{"id":"W4413963629","doi":"10.1016/j.jss.2025.112604","title":"Syntactic multilingual probing of pre-trained language models of code","year":2025,"lang":"en","type":"article","venue":"Journal of Systems and Software","topic":"Topic Modeling","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"Université de Montréal","funders":"Federación Española de Enfermedades Raras; Ministerio de Ciencia, Innovación y Universidades","keywords":"Computer science; Natural language processing; Code (set theory); Linguistics; Artificial intelligence; Programming language; Philosophy","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":[0.0005033996,0.00007354034,0.0003229622,0.0001437038,0.00002956577,0.00003388979,0.0002934277,0.00005179019,4.843668e-7],"category_scores_gemma":[0.0001360744,0.000057513,0.00006549808,0.0001151152,0.00001874615,0.0002832257,0.00007560982,0.0001135606,5.150848e-8],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00002256424,"about_ca_system_score_gemma":0.0001340444,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00009639433,"about_ca_topic_score_gemma":0.000005170198,"domain_scores_codex":[0.9989482,0.00004949764,0.0005841336,0.0001057454,0.0002129168,0.00009949504],"domain_scores_gemma":[0.9989318,0.0001440457,0.0004566168,0.0001988632,0.0002275693,0.00004107564],"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":[0.0003633683,0.0007944228,0.02592828,0.01647378,0.00129621,0.0003923566,0.1609469,0.4427409,0.06188358,0.06892172,0.0003729039,0.2198856],"study_design_scores_gemma":[0.001900295,0.0003026497,0.001771695,0.005707183,0.00008837084,0.0003003056,0.003815746,0.9762828,0.00706506,0.002421127,0.00009364435,0.0002510979],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.4574944,0.001759016,0.5403801,0.00003025487,0.0002157679,0.00007280744,0.00000225428,0.000009906264,0.00003550096],"genre_scores_gemma":[0.9734858,0.00001769791,0.02631674,0.00001073403,0.00003754833,0.000001010784,1.028289e-7,0.000003548509,0.0001268094],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.5335419,"threshold_uncertainty_score":0.2345312,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.01889991488062288,"score_gpt":0.2756434855286045,"score_spread":0.2567435706479816,"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."}}