{"id":"W7104817253","doi":"","title":"A Metamorphic Testing Perspective on Knowledge Distillation for Language Models of Code: Does the Student Deeply Mimic the Teacher?","year":2025,"lang":"","type":"article","venue":"ArXiv.org","topic":"Adversarial Robustness in Machine Learning","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Pipeline (software); Set (abstract data type); Perspective (graphical); Fidelity; Inference; Language model; Code (set theory); Software deployment","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":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002162069,0.0004388597,0.0005026103,0.0001745554,0.001136806,0.0001962915,0.002178644,0.0001404174,0.00001281668],"category_scores_gemma":[0.002527191,0.0002446023,0.0002869603,0.001236827,0.0004218223,0.0003716976,0.0009811869,0.0008694,0.0000133774],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.000439098,"about_ca_system_score_gemma":0.0003523114,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0003406505,"about_ca_topic_score_gemma":0.0001881952,"domain_scores_codex":[0.9965385,0.0008659922,0.0006835964,0.0009292612,0.0004496969,0.0005329752],"domain_scores_gemma":[0.9933532,0.003984647,0.0006062802,0.001337226,0.0006583009,0.00006032246],"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":[0.0002120772,0.000760727,0.1318689,0.0002234993,0.0008809273,0.000008642833,0.1331716,0.4918932,0.001395537,0.2039351,0.000133931,0.03551598],"study_design_scores_gemma":[0.0008273028,0.0001988983,0.08823535,0.0003086828,0.0003531724,0.000002352173,0.01319877,0.8902285,0.0004794203,0.005785115,0.00009611686,0.0002863951],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.5487404,0.001615791,0.4396797,0.004367204,0.001004526,0.001296035,0.0000115336,0.00008773778,0.003197099],"genre_scores_gemma":[0.9950457,0.000016483,0.002709597,0.000223931,0.0003692599,0.0001210083,0.00000219802,0.00003698949,0.001474806],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.4463053,"threshold_uncertainty_score":0.9974591,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.05337935037745778,"score_gpt":0.3541483723111379,"score_spread":0.3007690219336802,"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."}}