Skip to content
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

Return unclipped gradient from clip_grad_norm_ #756

Merged
merged 1 commit into from
Oct 14, 2022
Merged

Return unclipped gradient from clip_grad_norm_ #756

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
11 changes: 7 additions & 4 deletions src/accelerate/accelerator.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1194,6 +1194,9 @@ def clip_grad_norm_(self, parameters, max_norm, norm_type=2):
"""
Should be used in place of `torch.nn.utils.clip_grad_norm_`.

Returns:
`torch.Tensor`: Total norm of the parameter gradients (viewed as a single vector).

Example:

```python
Expand All @@ -1217,13 +1220,13 @@ def clip_grad_norm_(self, parameters, max_norm, norm_type=2):
parameters = [p for p in parameters]
for model in self._models:
if parameters == [p for p in model.parameters()]:
model.clip_grad_norm_(max_norm, norm_type)
return
return model.clip_grad_norm_(max_norm, norm_type)
elif self.distributed_type == DistributedType.DEEPSPEED:
# `accelerator.backward(loss)` is doing that automatically. Therefore, it's implementation is not needed
return
# We cannot return the gradient norm because DeepSpeed does it.
return None
self.unscale_gradients()
torch.nn.utils.clip_grad_norm_(parameters, max_norm, norm_type=norm_type)
return torch.nn.utils.clip_grad_norm_(parameters, max_norm, norm_type=norm_type)

def clip_grad_value_(self, parameters, clip_value):
"""
Expand Down