import numpy as np digits = load_digits() x = np.array(digits.data[:100]) y = np.array([[int(i == digit) for i in range(10)] for digit in digits.target[:100]]) validation_x = np.array(digits.data[100:120]) validation_y = np.array([[int(i == digit) for i in range(10)] for digit in digits.target[100:120]]) training_data = {'inputs': x, 'labels': y} validation_data = {'inputs': validation_x, 'labels': validation_y} P = Preprocessor.from_data(training_data) NN = NeuralNetwork.new([64, 10, 10], 'tanh') training_data = P.transform_data(training_data) validation_data = P.transform_data(validation_data) trainer = Trainer(NN, training_data, validation_data, classification=True) trainer.train({ 'learning_rate': 0.01, 'epoch_blocks': 10, 'batch_size': 100 }, { 'max_epochs': 2000, 'max_stall_blocks': 10 })