35b2c4917344f338eda67c78673cf4064b3b4265,examples/reinforcement_learning/tutorial_C51.py,DQN,train,#DQN#Any#Any#Any#Any#Any#,234

Before Change

        self._train_func(b_o, b_index, b_m)

        self.niter += 1
        if self.niter % target_q_update_freq == 0:
            sync(self.qnet, self.targetqnet)
            path = os.path.join(args.save_path, "{}.npz".format(self.niter))
            tl.files.save_npz(self.qnet.trainable_weights, name=path)

    @tf.function
    def _train_func(self, b_o, b_index, b_m):
        with tf.GradientTape() as tape:

After Change

        self.niter += 1
        if self.niter % target_q_update_freq == 0:
            sync(self.qnet, self.targetqnet)
            self.save(args.save_path)

    def save(self, path):
        if path is None:

Instances