{"id":"W3095672581","doi":"10.26434/chemrxiv.11869692","title":"RetroXpert: Decompose Retrosynthesis Prediction Like A Chemist","year":2020,"lang":"en","type":"preprint","venue":"ChemRxiv","topic":"Machine Learning in Materials Science","field":"Materials Science","cited_by":51,"is_retracted":false,"has_abstract":true,"ca_institutions":"NeuroDevNet","funders":"Cancer Prevention and Research Institute of Texas","keywords":"Retrosynthetic analysis; Synthon; Interpretability; Computer science; Artificial intelligence; Chemistry; Stereochemistry","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":["metaepi_narrow","insufficient_payload"],"consensus_categories":["insufficient_payload"],"category_scores_codex":[0.001158704,0.0005990976,0.0007834274,0.00009578228,0.0002646214,0.0006198192,0.001814647,0.0005458154,0.003813321],"category_scores_gemma":[0.001246933,0.0005833903,0.0002421525,0.0002443487,0.000337949,0.0001792294,0.001909191,0.0008723033,0.00126988],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0002758541,"about_ca_system_score_gemma":0.0002498841,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00009180576,"about_ca_topic_score_gemma":0.000002916806,"domain_scores_codex":[0.9957907,0.0002064534,0.000767965,0.001748088,0.0008451555,0.0006416421],"domain_scores_gemma":[0.9972984,0.0001402123,0.0006141602,0.00143885,0.0001540256,0.0003543053],"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":[0.00005466203,0.00004989041,0.0009132277,0.0004082173,0.00001275515,0.00002283621,0.0003525182,0.0004901544,0.9884406,0.00003565051,0.009090106,0.0001293885],"study_design_scores_gemma":[0.0002500002,0.00006113955,0.002579941,0.0003346285,0.00007418495,0.00002377594,0.00002888442,0.01333555,0.9720798,0.001351756,0.009146369,0.0007339653],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9813943,0.0002484148,0.00256702,0.002311773,0.004729081,0.0005212993,0.00008731472,0.001367558,0.006773258],"genre_scores_gemma":[0.9765248,0.00006429523,0.02038858,0.0005581994,0.001372597,0.0001567791,0.0001102673,0.0001050497,0.000719434],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.01782156,"threshold_uncertainty_score":0.9996617,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.02020156239878657,"score_gpt":0.260537496444046,"score_spread":0.2403359340452594,"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."}}