{"id":"W4281619372","doi":"10.1038/s41467-022-30839-x","title":"Language models can learn complex molecular distributions","year":2022,"lang":"en","type":"article","venue":"Nature Communications","topic":"Computational Drug Discovery Methods","field":"Computer Science","cited_by":176,"is_retracted":false,"has_abstract":true,"ca_institutions":"Canadian Institute for Advanced Research; Vector Institute; University of Toronto","funders":"","keywords":"Computer science; Generative grammar; Generative model; Artificial intelligence; Theoretical computer science; Graph; Molecular graph; Representation (politics); Language model; Machine learning; Natural language processing","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":[0.0004414509,0.0001192265,0.0001277562,0.0001331822,0.001006761,0.0001067976,0.004266364,0.00005635101,0.00003662935],"category_scores_gemma":[0.0001038572,0.0001385869,0.0001039491,0.001041082,0.00008431468,0.0002306574,0.00367682,0.001162628,0.000008650806],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0002141819,"about_ca_system_score_gemma":0.000198505,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00007101631,"about_ca_topic_score_gemma":0.00008529066,"domain_scores_codex":[0.9981277,0.0007645813,0.0002018936,0.0002787125,0.0004150252,0.0002120846],"domain_scores_gemma":[0.9966314,0.0004380419,0.00009277528,0.002636767,0.0001209763,0.00008002557],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.000001180878,0.0001225838,0.00002313489,0.00000150149,0.00001859231,0.000004753633,0.0007063341,0.06596898,0.0006039919,0.925063,0.002790337,0.00469566],"study_design_scores_gemma":[0.0002112219,0.00002528059,0.001434544,0.000002985673,0.00001243429,0.00004430063,0.0002454789,0.8556408,0.0001533518,0.05849624,0.0834869,0.0002464912],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.00774256,0.003736094,0.9338107,0.04044472,0.0002636056,0.0003455853,0.0004531722,0.0004091715,0.01279444],"genre_scores_gemma":[0.8322233,0.00001449256,0.1657255,0.001278134,0.00001116112,0.0001113305,0.0005347089,0.00001036579,0.00009102519],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.8665667,"threshold_uncertainty_score":0.7928036,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.04033398295059509,"score_gpt":0.3508870985748289,"score_spread":0.3105531156242338,"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."}}