Shashank Rammoorthy
Shashank Rammoorthy
I'm having the exact same issue - but deleting the lockfile doesn't seem to fix it. Any pointers would be super appreciated! @skye @ayaka14732
@skye I can verify I still see the message even after I delete both `tpu_logs` and the lockfile. I've been running processes as Python subprocesses through a script - is...
Is this still available? I'd be interested!
but we'd have to think about how to best design this - maybe there's a client side operator map as well - this could generalize well for webgpu stuff too
Hmm so I still see this same issue @kptkin - I have `accelerate` version `1.0.1` and `wandb` version `0.18.3`. Even running `wandb.init()` leads to the same issue (`wandb.errors.errors.CommError: Run initialization...
same issue - is this resolved with the LiteLLM integration at all?