Skip to content

GlobalNormClip use inplace mul #31686

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

Closed
wants to merge 3 commits into from
Closed

GlobalNormClip use inplace mul #31686

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
7615517
GlobalNormClip use inplace mul
wangxicoding Mar 17, 2021
4447d2e
fix var in c_reduce_sum
wangxicoding Mar 17, 2021
9d66415
fix g
wangxicoding Mar 19, 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
3 changes: 2 additions & 1 deletion python/paddle/distributed/fleet/meta_optimizers/sharding/gradient_clip_helper.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -48,7 +48,8 @@ def prune_gradient_clip(self, block, shard):
if deperate_op:
deperate_op_idx.add(idx)
for output_name in op.desc.output_arg_names():
deperated_vars.add(output_name)
if output_name not in op.desc.input_arg_names():
deperated_vars.add(output_name)

if not deperated_vars:
# got no gradient_clip op
Expand Down
11 changes: 8 additions & 3 deletions python/paddle/fluid/clip.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -489,9 +489,14 @@ def _static_clip(self, params_grads):
continue

with p.block.program._optimized_guard([p, g]):
new_grad = layers.elementwise_mul(x=g, y=scale_var)
param_new_grad_name_dict[p.name] = new_grad.name
params_and_grads.append((p, new_grad))
p.block.append_op(
type='elementwise_mul',
inputs={'X': g,
'Y': scale_var},
outputs={'Out': g})

param_new_grad_name_dict[p.name] = g.name
params_and_grads.append((p, g))

_correct_clip_op_role_var(params_and_grads, param_new_grad_name_dict)
return params_and_grads
Expand Down