Sanyam Kapoor

Here's a tiny hack meanwhile. I modified `viewer_setup()` method in [`mujoco_env.py`](https://github.com/rlworkgroup/metaworld/blob/master/metaworld/envs/mujoco/mujoco_env.py#L77) with the following statements ```python ... def viewer_setup(self): self.viewer.cam.azimuth = -20 self.viewer.cam.elevation = -20 ... ``` Gets you this...

Here is log output: ``` Starting FlowVisor Setting debugging mode: all logs to stderr --- Setting logging level to NOTE CRIT:2015-03-03T19:42:07.009:none:: log level enabled: CRIT ALERT:2015-03-03T19:42:07.048:none:: log level enabled: ALERT...

Warning: This is a trace based export. Some side-effects might be unintended.

Need this for Backplay.

Thanks for the info @fehiepsi @neerajprad The issue seems to still exist with `mp_context='forkserver'`.

Hey @martinjankowiak, allow me to suggest something. I'm sure the Bayesian literature has collected plenty of tutorial data sets and tasks over the years (think counterparts for MNIST/CIFAR/COCO from the...

Just wanted to chime in with another use case. I currently schedule jobs on a machine with GPUs using `pueue`. Most times, I need multiple GPUs and set the `CUDA_VISIBLE_DEVICES`...

Thank you. My understanding was that this behavior would require me to change existing code to parse out the `PUEUE_WORKER_ID` and transform it to appropriate values of `CUDA_VISIBLE_DEVICES` (e.g. something...

Ahaa! That makes much more sense. Thank you for clarifying. Let me run a quick test simulating this scenario, and will report back. I'll update the wiki with this example...

Sorry for the delay. I have verified this behavior, and this actually solves my specific problem. I also realized I missed this: > There'll never be two concurrent tasks with...