tu

ok @lespeholt I will look into it again on running _without_ docker. Also, is it possible to use grpc python instead of grpc c++? do you know if there would...

@lespeholt > frames: [time, batch_size, height, width, channels]. These should > be un-normalized frames in range [0, 255]. channels must be equal to 1 > when we actually stack frames...

In an effort to make the stacking code more generic to accommodate other environments, I've modified the frame stacking code to the following: `STACKING_STATE_DTYPE = tf.float32` ``` def initial_frame_stacking_state(stack_size, batch_size,...

hi @lespeholt I've taken a look at networks_test.py and was wondering why [input frame](https://github.com/google-research/seed_rl/blob/eff7aaa7ab5843547fbf383fcc747b7a8ca67632/atari/networks_test.py#L187) `frames=[[[1]]]` in `test_stack_frames(self)` has shape `[time=1, batch_size=1, channels=1]` used for testing? ``` def test_stack_frames(self): zero_state =...

I seem to be getting an error when I run `networks_test.py`'s `test_stack_frames()` and `test_stack_frames_done()` functions with my converted `networks.py` code at: ``` extended_frames = tf.concat( [tf.reshape(frame, [1] + frame.shape +...

From the example, `Dmat` has shape `(2, 8)` and `dvec` has shape `(2,)`. What does the `2` dimension represent? Also, its not immediately clear to me how to write out...

Thanks for the detailed explanation! > Have more samples than variables (assets) 1. By samples, do you mean batch size? > Do not compute the square root at all (skip...

Okay that makes sense. Lastly since we are on the topic of `CovarianceMatrix`, I have two questions regarding it being implemented in `Resample`. 1. Based on previous discussion, > Otherwise...

hi @jankrepl , is there a rule of thumb for this? Like what is the typical ratio of lookback to assets. > Have more samples than variables (assets)

Hi @jankrepl , I have stumbled upon a promising covariance estimation method that would do better when `lookback` < `n_assets` compared to the current sample covariance method, called [T-CorEx](https://arxiv.org/pdf/1905.13276v2.pdf). Compared...