{"id":"W4403856538","doi":"10.1101/2024.10.25.620213","title":"Towards Robust Evaluation of Protein Generative Models: A Systematic Analysis of Metrics","year":2024,"lang":"en","type":"preprint","venue":"bioRxiv (Cold Spring Harbor Laboratory)","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":1,"is_retracted":false,"has_abstract":true,"ca_institutions":"Air Canada","funders":"","keywords":"Generative grammar; Computer science; Artificial intelligence","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":"metacan-v3-hybrid-931329e0061c","candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.06538694,0.003522737,0.002962593,0.007713547,0.001324877,0.005928942,0.003241413,0.002718749,0.001128506],"category_scores_gemma":[0.2465399,0.001058833,0.001738441,0.005095298,0.003750256,0.006611135,0.005993483,0.003530693,0.0004991095],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.004419597,"about_ca_system_score_gemma":0.004261139,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002686738,"about_ca_topic_score_gemma":0.002490734,"domain_scores_codex":[0.9440522,0.03414836,0.004725449,0.004047615,0.01226604,0.00076033],"domain_scores_gemma":[0.8047671,0.1419807,0.01136453,0.02190175,0.01855054,0.001435415],"domain_codex":null,"domain_gemma":"evaluation","domain_candidate":"evaluation","domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"systematic_review","study_design_scores_codex":[0.0006843085,0.0005150806,0.02988156,0.003874243,0.001280625,0.0003249332,0.0008669415,0.5794211,0.0146374,0.0713242,0.01087085,0.2863188],"study_design_scores_gemma":[0.00006655662,0.0006439564,0.003567968,0.0005261193,0.000137515,0.0002856411,0.000226213,0.9225549,0.01206028,0.05368236,0.006149371,0.00009915073],"study_design_candidate":"systematic_review","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.08894888,0.01008771,0.890198,0.001653242,0.0002420988,0.0008707922,0.001524062,0.003062528,0.003412684],"genre_scores_gemma":[0.3926496,0.002821403,0.5964856,0.0005061655,0.0001485897,0.001221015,0.004163904,0.001436375,0.0005673195],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.934613,"threshold_uncertainty_score":0.3458033,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.03180507768929152,"score_gpt":0.2471167065462327,"score_spread":0.2153116288569412,"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."}}