{"id":"W2903425689","doi":"10.3389/fphar.2020.565644","title":"Molecular Sets (MOSES): A Benchmarking Platform for Molecular Generation Models","year":2020,"lang":"en","type":"preprint","venue":"Frontiers in Pharmacology","topic":"Machine Learning in Materials Science","field":"Materials Science","cited_by":84,"is_retracted":false,"has_abstract":true,"ca_institutions":"Vector Institute; University of Toronto; Canadian Institute for Advanced Research","funders":"","keywords":"Benchmarking; Generative grammar; Computer science; Set (abstract data type); Machine learning; Generative model; Code (set theory); Training set; Quality (philosophy); Chemical space; Artificial intelligence; Data mining; Bioinformatics; Programming language; Biology","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":[],"consensus_categories":[],"category_scores_codex":[0.004201765,0.002228895,0.001273512,0.001784444,0.0008595679,0.00186151,0.005021903,0.002541983,0.0209284],"category_scores_gemma":[0.01033953,0.0009343695,0.002113508,0.001763009,0.000857425,0.002327694,0.002538834,0.002961036,0.007670668],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001377833,"about_ca_system_score_gemma":0.002139047,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.004066108,"about_ca_topic_score_gemma":0.004705303,"domain_scores_codex":[0.9979761,0.0006987599,0.0001588283,0.0002748288,0.0007381784,0.0001533264],"domain_scores_gemma":[0.9971842,0.001487849,0.0001564185,0.0006420868,0.0003655271,0.0001638553],"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":[0.0009971192,0.0008643102,0.006145048,0.002691033,0.0008502224,0.0004302636,0.0002242007,0.585887,0.01057447,0.07438264,0.1892336,0.12772],"study_design_scores_gemma":[0.0003393869,0.0004139564,0.001054383,0.0001655941,0.00007736721,0.0001707521,0.00004779474,0.8751385,0.01353008,0.02924203,0.0797307,0.00008950445],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.08390774,0.007770672,0.5642421,0.002872573,0.001496018,0.001404539,0.06451819,0.2172107,0.0565775],"genre_scores_gemma":[0.3399761,0.004354655,0.459518,0.001396526,0.0002462017,0.003058588,0.1476804,0.03191024,0.01185922],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.0209284,"threshold_uncertainty_score":0.07001251,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.03716816222312058,"score_gpt":0.3199947715576369,"score_spread":0.2828266093345164,"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."}}