uzhilinsky
uzhilinsky
We've decided to implement grad accumulation. Please stay tuned.
Just to confirm, is upgrading from 2.10.4 to 2.10.6 is what solving this for you? (2.10.4 is what we currently install). @Raziel90 , if you are still running into this,...
Hard to say without more context. I would start by looking at the action distribution difference between the two datasets. A quick sanity check might be to just look at...
We've certainly been able to successfully fine-tune pi0-base on novel tasks. Hard to say what's going on in your case without having more context. Just in case, are you using...
100,000 steps is a lot for such a small dataset. It's very likely that the model has overfit to your data. Note that we are using 20K or 30K in...
Have you seen the Docker [docs](https://github.com/Physical-Intelligence/openpi/blob/main/docs/docker.md)? We are listing a bunch of limitations there.
Have you tried [running with Docker](https://github.com/Physical-Intelligence/openpi/blob/main/examples/libero/README.md#with-docker)? If it works, JAX is likely picking up the wrong cuda dependencies.
This was addressed at HEAD.