Nazar

JavaCPP requires a lot of handwork. In most cases it is difficult to generate suitable Java interfaces. I have used JavaCPP for PyTorch api, you may find it here: https://github.com/nazarblch/torch-scala

I had not solved this problem that time. Samplers actually do not affect the speed too much in algorithms with buffers.

if your problem relates to dreamer algorithm consider also the other implementations https://github.com/yusukeurakami/dreamer-pytorch https://github.com/ray-project/ray/blob/master/rllib/agents/dreamer/dreamer.py

Yes, you can build it with https://download.pytorch.org/libtorch/cpu/libtorch-shared-with-deps-latest.zip Pull from the repo first and don't forget to export JAVA_HOME I would propose to discuss this project in skype, what do you...

> @nazarblch thanks a lot for the contribution! Still need to finish looking at `_loss_fn` and `_euclidean_loss_fn`'s implementation + the notebook, but left other comments. This is a short confirmation...

@michalk8, I have corrected the code and updated the repository. Also we uploaded a new version of the paper into arxiv with additional image to image translation experiments.