Regarding the issue of the model crashing during training #65
Description
When I adjust the parameters to num_steps = 150 and no_im_loss_steps = 20, my model crashes and cannot encode as soon as the program executes _, _, global_step = sess.run([train_op, loss_op, global_step_tensor], feed_dict=feed_dict). I have also attached an image.
