) for t in xrange(epoch_cycle): print 'epoch', t current_cost = 0 #validate_cost = 0 x_batch, y_hat_batch, batch_num = DataParser.make_batch(batch_size) #x_batch = [[[-2.50, -1.39], [-2.50, 0.92]], [[-2.50, -1.39], [-2.49, 0.93]], [[-2.50, -1.39], [-2.48, 0.94]]] #y_hat_batch = [[[0,1]], [[0,1]], [[0,1]]] for i in xrange(batch_num): c_cost = train(x_batch[i], y_hat_batch[i]) #print 'c_cost', c_cost current_cost += c_cost current_cost /= batch_num print 'current_cost', current_cost DataParser.save_parameters(parameters) #current_validate_cost = validate(validation_set) #print 'validate cost', current_validate_cost #if current_validate_cost < validate_cost: # validate_cost = current_validate_cost #else: # break test_data, test_id = DataParser.load_test_data(sample_file) result = test(test_data) result = list(result) result = map(list, zip(*result)) # transpose with open('result.txt', 'w') as f: for i in xrange(len(result)): f.write(test_id[i] + ',') max_value = 0
) for t in xrange(epoch_cycle): print 'epoch', t current_cost = 0 #validate_cost = 0 x_batch, y_hat_batch, batch_num = DataParser.make_batch(batch_size) #x_batch = [[[-2.50, -1.39], [-2.50, 0.92]], [[-2.50, -1.39], [-2.49, 0.93]], [[-2.50, -1.39], [-2.48, 0.94]]] #y_hat_batch = [[[0,1]], [[0,1]], [[0,1]]] for i in xrange(batch_num): c_cost = train(x_batch[i], y_hat_batch[i]) #print 'c_cost', c_cost current_cost += c_cost current_cost /= batch_num print 'current_cost', current_cost DataParser.save_parameters(parameters, 'parameter.txt') DataParser.save_parameters(grad_hists, 'parameter-g.txt') #current_validate_cost = validate(validation_set) #print 'validate cost', current_validate_cost #if current_validate_cost < validate_cost: # validate_cost = current_validate_cost #else: # break test_data, test_id = DataParser.load_test_data(sample_file) result = test(test_data) result = list(result) result = map(list, zip(*result)) # transpose with open('result.txt', 'w') as f: for i in xrange(len(result)): f.write(test_id[i] + ',')