matthewdeng
matthewdeng
@amogkam @clarkzinzow @bveeramani assigning to you all to review the API!
@YQ-Wang can you share what you're running and what the error you're getting is?
@YQ-Wang yes `ray` should be fully compatible with `numpy` 1.24, which is the version contains in the Docker images. As always, it is still possible that other dependencies maybe introduce...
@aslonnie can you resolve conflicts on this one?
Can you share the full stack trace? This might be due to other dependencies (perhaps getting pulled in as extras).
Hi @kamal-rahimi, Actually the default behavior should be to keep the last checkpoint, for the exact reasons you've mentioned. Are you observing something different? **Example:** ```python import ray from ray.train.torch...
@Yard1 would aliasing the class work?
@MahdiNazemi can you take a look at https://github.com/ray-project/ray/issues/31451#issuecomment-1371590235 and see if this is applicable to your script? Specifically, do you use a `DataLoader` with `num_workers>0`?
Ah okay I think that's likely it - please try with `num_workers=0` as I believe `num_workers=1` will still end up launching 1 subprocess, which could run into the same issue.