-
Notifications
You must be signed in to change notification settings - Fork 5.4k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
ray.util.collective support torch.bfloat16 #39845
Merged
stephanie-wang
merged 2 commits into
ray-project:master
from
wuxibin89:feat/collectvie_bfloat16
May 15, 2024
Merged
ray.util.collective support torch.bfloat16 #39845
stephanie-wang
merged 2 commits into
ray-project:master
from
wuxibin89:feat/collectvie_bfloat16
May 15, 2024
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Signed-off-by: wuxibin <wuxibin89@163.com>
jackhumphries
approved these changes
Mar 25, 2024
@stephanie-wang can you merge? |
Thanks for the contribution, @wuxibin89 ! |
ryanaoleary
pushed a commit
to ryanaoleary/ray
that referenced
this pull request
Jun 6, 2024
[bfloat16](https://en.wikipedia.org/wiki/Bfloat16_floating-point_format) is widely used in LLM training and inference since it can achieve higher throughput and is less prone to weight growth. ray.util.collective use cupy.cuda.nccl for GPU communication, while cupy doesn't support bfloat16 for now (cupy/cupy#7527). So for allgather/reducescater operation, we should bypass cupy.array and use torch.tensor directly. Signed-off-by: wuxibin <wuxibin89@163.com> Co-authored-by: Stephanie Wang <swang@cs.berkeley.edu> Signed-off-by: Ryan O'Leary <ryanaoleary@google.com>
ryanaoleary
pushed a commit
to ryanaoleary/ray
that referenced
this pull request
Jun 6, 2024
[bfloat16](https://en.wikipedia.org/wiki/Bfloat16_floating-point_format) is widely used in LLM training and inference since it can achieve higher throughput and is less prone to weight growth. ray.util.collective use cupy.cuda.nccl for GPU communication, while cupy doesn't support bfloat16 for now (cupy/cupy#7527). So for allgather/reducescater operation, we should bypass cupy.array and use torch.tensor directly. Signed-off-by: wuxibin <wuxibin89@163.com> Co-authored-by: Stephanie Wang <swang@cs.berkeley.edu> Signed-off-by: Ryan O'Leary <ryanaoleary@google.com>
ryanaoleary
pushed a commit
to ryanaoleary/ray
that referenced
this pull request
Jun 7, 2024
[bfloat16](https://en.wikipedia.org/wiki/Bfloat16_floating-point_format) is widely used in LLM training and inference since it can achieve higher throughput and is less prone to weight growth. ray.util.collective use cupy.cuda.nccl for GPU communication, while cupy doesn't support bfloat16 for now (cupy/cupy#7527). So for allgather/reducescater operation, we should bypass cupy.array and use torch.tensor directly. Signed-off-by: wuxibin <wuxibin89@163.com> Co-authored-by: Stephanie Wang <swang@cs.berkeley.edu>
GabeChurch
pushed a commit
to GabeChurch/ray
that referenced
this pull request
Jun 11, 2024
[bfloat16](https://en.wikipedia.org/wiki/Bfloat16_floating-point_format) is widely used in LLM training and inference since it can achieve higher throughput and is less prone to weight growth. ray.util.collective use cupy.cuda.nccl for GPU communication, while cupy doesn't support bfloat16 for now (cupy/cupy#7527). So for allgather/reducescater operation, we should bypass cupy.array and use torch.tensor directly. Signed-off-by: wuxibin <wuxibin89@163.com> Co-authored-by: Stephanie Wang <swang@cs.berkeley.edu> Signed-off-by: gchurch <gabe1church@gmail.com>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Why are these changes needed?
bfloat16 is widely used in LLM training and inference since it can achieve higher throughput and is less prone to weight growth. ray.util.collective use cupy.cuda.nccl for GPU communication, while cupy doesn't support bfloat16 for now (cupy/cupy#7527). So for allgather/reducescater operation, we should bypass cupy.array and use torch.tensor directly.
Related issue number
Checks
git commit -s
) in this PR.scripts/format.sh
to lint the changes in this PR.method in Tune, I've added it in
doc/source/tune/api/
under thecorresponding
.rst
file.