def calculate_predictions(experiment_name, dataset): print('Loading data... ') sys.stdout.flush() if experiment_name in ['danq', 'deepsea', 'danqjaspar']: data = Data(data_suffix='_full') X, y = data.get_data(dataset) else: data = get_data_loader(experiment_name) X, y = data.get_data(dataset) print('Loading model... ') sys.stdout.flush() model = get_trained_model(experiment_name) print('Calculating predictions... ') sys.stdout.flush() make_predictions(model, X, join( RESULT_DIR, 'predictions-best', '{}-{}{}.npy'.format(experiment_name, dataset, data.suffix)), verbose=1)
import os, sys from core.data import Data from core.train_model import get_trained_model def append_to_losses(expt_name, dataset, loss, filename='final_losses_{}.csv'.format(sys.argv[2])): with open(filename, 'a') as f: f.write('{},{},{}\n'.format(expt_name, dataset, loss)) RESULT_DIR = os.environ.get('RESULT_DIR', 'results') data = Data(sequence_length=int(sys.argv[2]), data_suffix='_full') m = get_trained_model(sys.argv[1]) print('evaluating model', flush=True) l = m.evaluate(*data.get_data('test')) print('saving results', flush=True) append_to_losses(sys.argv[1], 'test', l)