q3_gru.py 文件源码

def add_training_op(self, loss):
        """Sets up the training Ops.

        Creates an optimizer and applies the gradients to all trainable variables.
        The Op returned by this function is what must be passed to the
        `sess.run()` call to cause the model to train. See

        TODO:
            - Get the gradients for the loss from optimizer using
              optimizer.compute_gradients.
            - if self.clip_gradients is true, clip the global norm of
              the gradients using tf.clip_by_global_norm to self.config.max_grad_norm
            - Compute the resultant global norm of the gradients using
              tf.global_norm and save this global norm in self.grad_norm.
            - Finally, actually create the training operation by calling
              optimizer.apply_gradients.
        See: https://www.tensorflow.org/api_docs/python/train/gradient_clipping
        Args:
            loss: Loss tensor.
        Returns:
            train_op: The Op for training.
        """

        optimizer = tf.train.GradientDescentOptimizer(learning_rate=self.config.lr)

        ### YOUR CODE HERE (~6-10 lines)

        # - Remember to clip gradients only if self.config.clip_gradients
        # is True.
        # - Remember to set self.grad_norm

        ### END YOUR CODE

        assert self.grad_norm is not None, "grad_norm was not set properly!"
        return train_op