{"id":"W4231139425","doi":"10.26434/chemrxiv.14696595","title":"Inside the Black Box: A Physical Basis for the Effectiveness of Deep Generative Models of Amorphous Materials","year":2021,"lang":"en","type":"preprint","venue":"ChemRxiv","topic":"Machine Learning in Materials Science","field":"Materials Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"McGill University","funders":"Natural Sciences and Engineering Research Council of Canada; Compute Canada","keywords":"Amorphous solid; Autoregressive model; Generative grammar; Protocol (science); Scale (ratio); Sampling (signal processing); Computer science; Black box; Artificial intelligence; Algorithm; Mathematics; Chemistry; Econometrics; Medicine; Cartography; Computer vision; Geography","routes":{"ca_aff":true,"ca_fund":true,"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":[0.004487844,0.0004406711,0.001117425,0.00005371569,0.000197244,0.0002700602,0.001703199,0.00020785,0.0002471758],"category_scores_gemma":[0.001386513,0.0002557651,0.0003154654,0.0002035369,0.001172799,0.0001445063,0.001473597,0.0002760222,0.000009860967],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00007892317,"about_ca_system_score_gemma":0.0002808534,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0004104187,"about_ca_topic_score_gemma":0.00001609022,"domain_scores_codex":[0.9959715,0.001500465,0.0006839777,0.0008396829,0.0005890129,0.0004153644],"domain_scores_gemma":[0.9937301,0.003299877,0.0009181612,0.0014753,0.0005185367,0.0000579906],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","study_design_scores_codex":[0.0002009468,0.00009310741,0.00001810831,0.001158615,0.00005370541,0.000001715099,0.002005412,0.1472703,0.8478607,0.001266303,0.00002256441,0.00004854973],"study_design_scores_gemma":[0.000242399,0.00006651517,0.0009812933,0.0003475888,0.0001526791,0.000003747658,0.0002143791,0.03763586,0.9525953,0.007494437,0.00000762622,0.0002581554],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9468266,0.0001952889,0.04977214,0.0001540641,0.001250578,0.001483442,0.00009600467,0.00004552091,0.0001762981],"genre_scores_gemma":[0.9928573,0.00003403311,0.006011281,0.0000599162,0.0003382133,0.0005951442,0.00002679733,0.00005526108,0.0000220038],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.1096344,"threshold_uncertainty_score":0.9999894,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.0241197385242391,"score_gpt":0.2843745611565126,"score_spread":0.2602548226322735,"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."}}