from neural_network import neural_network import numpy as np import tensorflow as tf from neural_network.network_structure.Logger import Logger import os network = neural_network.neural_network() logger = Logger(network.path) train_data, train_labels, test_data, test_labels = network.data() EPOCHS = 100 BATCH_SIZE = 32 def log_write(data): with open(os.path.join(network.path, 'log.txt'), 'a') as file: file.write(data + '\n') print(data) def train(): sess = tf.Session() network.create_model() network.session_init(sess) for epoch in range(EPOCHS): log_write('##########%4d##########' % (epoch)) train_loss, train_accuracy = train_one_epoch(train_data, train_labels, epoch) test_accuracy = test_one_epoch(test_data, test_labels) logger.log_scalar(tag='Loss per Epoch', value=train_loss, step=epoch) logger.log_scalar(tag='Train Accuracy per Epoch',
import tensorflow as tf import os import numpy as np from neural_network.neural_network import neural_network from neural_network.network_structure.Logger import Logger from IPython import embed import argparse import sys parser = argparse.ArgumentParser() parser.add_argument('--mode',type=str,default='train',help='Mode of operation') parser.add_argument('--weights',type=str,default='1.ckpt',help='Path of weights') args = parser.parse_args() # Create a Neural Network Class. nn = neural_network() # Save network structure in logs. nn.save_network_structure() L1,L2 = 1,1 def end_effector_pose(t1,t2): return [L1*np.cos(t1)+L2*np.cos(t1+t2),L1*np.sin(t1)+L2*np.sin(t1+t2)] # Create logger file for tensorboard. # Get the path from neural network class. logger = Logger(nn.path) episodes = 5000 batch_size = 100 samples = 10000