return-sleep

hi, do you solve this problem? i take the error too.

b = tf.zeros(shape=[tf.shape(inputs_hat)[0], self.num_capsule, self.input_num_capsule, 1, 1]) some wrong happened. NotImplementedError: Cannot convert a symbolic Tensor (digitcaps/strided_slice:0) to a numpy array.

you should sync the submodule: git submodule update --recursive --init or download related folder from https://github.com/dccastro/Morpho-MNIST.git @woqingdoua

Thank you for your reply, when I changed the version of hickle, it worked.

@TheSunWillRise Can you please share the training logs on the minist dataset, e.g. loss at training convergence. I've been trying this experiment recently, but found that the prediction loss oscillates...

> During my training, usually it is stable especially for mojoco tasks. Which environment are you testing? If it happens, maybe you can moniter the q-value function loss. Usually decreasing...

Hi, I've been bothered by the stochasticity of D4RL gym tasks. Can we have a further discussion?

Have you solved the problem yet?

> import gym import d4rl env_name = "halfcheetah-medium-v2" env = gym.make(env_name) dataset = d4rl.qlearning_dataset(env) > > I have the same issue with the above code The problem may be caused...

I do not quite understand the `_shift_sequences` function here， why we should shift the transition sequences.?