Skip to content

[XPU] hybrid_parallel_optimizer xpu support bf16 #60213

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.

Sign up for GitHub

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

Merged
houj04 merged 1 commit into from
Dec 21, 2023
Merged

[XPU] hybrid_parallel_optimizer xpu support bf16 #60213

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
12 changes: 2 additions & 10 deletions ...n/paddle/distributed/fleet/meta_optimizers/dygraph_optimizer/hybrid_parallel_optimizer.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -233,12 +233,8 @@ def _comm_and_clip(
)
clip_var_fp16 = paddle.cast(clip_var, paddle.float16)

# bf16 is not supported on XPU now
if not (
paddle.is_compiled_with_xpu()
or isinstance(
paddle.framework._current_expected_place(), paddle.CustomPlace
)
if not isinstance(
paddle.framework._current_expected_place(), paddle.CustomPlace
):
clip_var_bf16 = paddle.cast(clip_var, paddle.bfloat16)
for p, g in params_grads:
Expand All @@ -249,10 +245,6 @@ def _comm_and_clip(
if g.dtype == paddle.float16:
g.multiply_(clip_var_fp16)
elif g.dtype == paddle.bfloat16:
if paddle.is_compiled_with_xpu():
raise NotImplementedError(
"BF16 is not supported on XPU now"
)
g.multiply_(clip_var_bf16)
else:
g.multiply_(clip_var)
Expand Down