if __name__ == '__main__': from train import standard_arg_paser, generate_train_cmd, run_cmds, defaults, all_models, all_output_types parser = standard_arg_paser() args = parser.parse_args() # Settings checkpoint_base_dir = "/scratch1/rwt891/data/deepfold/camara" test_base_dir = "/scratch1/rwt891/data/deepfold/camara_test" id_number = "01" # Generate commands cmds = [] data_dir_base = "/scratch1/rwt891/data/cull_pdb_pc100_entries_170602/culled_pc30_res3.0_R0.3_d170611/" data_name = "pc30" num_passes = "10" for model in [ "CubedSphereModel", "SphericalModel", "CartesianHighresModel" ]: for output_type in all_output_types: cmd = generate_train_cmd( args.mode, model, output_type, data_name, defaults['regularization'], defaults['learning_rate'],
'--model', model ] if restore_checkpoint == 'best': best = get_best_checkpoint(checkpoint_base_dir + "/" + name + ".txt") cmd += ["--step", best] output_file = output_dir + '/' + ddg_name + '_' + restore_checkpoint + "_" + name + '.txt' return cmd, output_file, name if __name__ == '__main__': from train import standard_arg_paser, run_cmds, defaults, all_models, all_output_types parser = standard_arg_paser(exclude=['mode']) args = parser.parse_args() # Settings checkpoint_base_dir = "/scratch1/rwt891/data/deepfold/camara" output_dir = "/scratch1/rwt891/data/deepfold/camara_ddg" ddg_data_dir_base = "/scratch1/rwt891/data/ddgs" train_data_name = "pc30" train_num_passes = "10" train_id_number = "01" output_type = 'aa' # Generate commands