Open amineoui opened 10 months ago
Hi amineoui, sorry for the delayed response. The problem is that I used resnet modules(e.g. ResNetRepresentation ) instead of MLP ones(Representation
in your code) to process image input. And here is an example project that uses muax and accepts visual inputs: NTS
i try MCTX Muax its a good art work , but i have issue when i set conv observation like a shape (96, 96, 1) , please guide me to solve this problem and the correct way to set train_env and eval_env as input of muax.fit() function
i use this code :
a simple example how i can run the muax with atari gym PongNoFrameskip-v4