Skip to content

[hybrid] [npu] fit npu nan/inf check #35171

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
merged 2 commits into from
Sep 2, 2021
Merged

[hybrid] [npu] fit npu nan/inf check #35171

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
77112f8
[hybrid] fit npu nan/inf check
FeixLiu Aug 26, 2021
e9f4e2f
add comment
FeixLiu Aug 27, 2021
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
8 changes: 7 additions & 1 deletion python/paddle/fluid/optimizer.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -5316,7 +5316,13 @@ def _accumulate_gradients_with_fuse(self, main_block, fp16, fused_size):
"copy_data": False,
"use_align": True,
"dtype": grads[0].dtype,
self._op_role_key: self._op_role.Backward
self._op_role_key: self._op_role.Backward,
# On npu, the nan/inf check login is different with gpu.
# If there are some not initialized sections in the fused var,
# and the value in those sections are nan/inf, it will trigger the nan/inf check.
# To avoid these problematic triggers, set constant is needed for npu
"set_constant": core.is_compiled_with_npu(),
"constant": float(0.0),
})
offset += 1
# For the gradient_merged_fused_var, given a init value during the coalesce op
Expand Down