albanD
albanD
@pytorchbot merge @namgyu-youn avoid force merging when no test actually ran yet ;)
@pytorchbot merge
Sounds good!
@pytorchbot revert -m "Broken trunk distributed ci" -c ignoredsignal Still breaks distributed tests: https://github.com/pytorch/pytorch/actions/runs/11171992932/job/31059959841 so maybe more skips are needed until the upgrade is done?
I think https://pytorch.org/docs/stable/notes/autograd.html#locally-disabling-gradient-computation contains the most in-depth discussion of this topic. I'm sure we could expand on it in a tutorial if there are enough questions we want to cover....
@XuehaiPan @ezyang doesn't this new recommendation break ccache due to folder randomization? Why do we recommend that??
But then I'm surprised of the very big warning that was added to vllm recently: https://docs.vllm.ai/en/latest/getting_started/installation/gpu.html#full-build-with-compilation ?
EDIT: From @colesbury this is due to https://github.com/python/cpython/commit/06d76c4b94f87b0785a96dbfa0904afc0b459cb7 on the cpython side that changes the ABI by adding a field. So this is expected to segfault until we re-build a...
Public API changes look good now!
Hey! This is expected as this tutorial is for the 2.6 release that will come out a bit later this week ! @svekars I think it's ok to keep this...