when I run this script without tune, I got “tensorflow.python.framework.errors_impl.InvalidArgumentError: Exception encountered when calling layer ‘relative_multi_head_attention’ (type RelativeMultiHeadAttention).” It happend after finished training and start running test loop.
what’s that mean? or how to fix it?
I’m using MBA with M1 apple silicon.
python: 3.10.10
ray: 2.3.1
tensorflow-macos: 2.12.0
well, I have fixed it!
Excellent. We will close this issue, then.
@JesseLT how did u solve this? i am getting an error Tensors must have same number of dimensions: got 2 and 3 when running algo.compute_single_action(obs, state)
have a try to add the code: state = np.expend_dims(state, axis=0) after the line: state = algo.get_policy().get_initial_state()