Johnny He comments

Results 5 comments of


                                            Johnny He

Loss function in SAC

> The equation is for the derivative of J_pi, but we needn't calculate the derivative explicitly - PyTorch does it for us. Sorry about that, I have re-edited the problem...

Won't run on school's compute

I found that if I run my scripts locally, I would not get this error. But when I visit my computer via ssh tool, this happens. So I doubt that...

Couldn't reproduce the result on MuJoCo Suite (d4rl datasets).

Great! Thanks for your warm reply! I will try to implement the BEAR results again. For the d4rl datasets, I use this [version](https://github.com/rail-berkeley/d4rl/commit/4730e689db9677389f1ecaa9fc8ad76414ea82d8) (20200803).

QF_Loss backprops policy network

@dssrgu Thanks for your contribution!, I have adapted your commits. But here are some questions. Does your modification works? Have you compared with the original version (PyTorch version==1.4, work)? If...

Cannot Share GPUs between processes when using JAX with SLURM

Similar issue. Waiting for a solution... I cannot run multiple GPUs in a GPU.