Skip to content

GitLab

"tests/vscode:/vscode.git/clone" did not exist on "e7602a4c47d7befd8177bc51cdb40a5794df982c"
Commit a4afb428 authored by Lawrence McAfee's avatar Lawrence McAfee
Browse files

cleaned clip_grads.py.

parent 787882a0
Show whitespace changes
Inline Side-by-side
Showing with 2 additions and 3 deletions
megatron/optimizer/clip_grads.py
View file @ a4afb428
...@@ -26,8 +26,7 @@ from megatron.mpu.layers import param_is_not_tensor_parallel_duplicate ...@@ -26,8 +26,7 @@ from megatron.mpu.layers import param_is_not_tensor_parallel_duplicate
def clip_grad_norm_fp32(parameters, max_norm, norm_type=2, def clip_grad_norm_fp32(parameters, max_norm, norm_type=2,
model_parallel_group=None, model_parallel_group=None):
ITERATION=None):
"""Clips gradient norm of an iterable of parameters whose gradients """Clips gradient norm of an iterable of parameters whose gradients
are in fp32. are in fp32.
...@@ -41,7 +40,7 @@ def clip_grad_norm_fp32(parameters, max_norm, norm_type=2, ...@@ -41,7 +40,7 @@ def clip_grad_norm_fp32(parameters, max_norm, norm_type=2,
max_norm (float or int): max norm of the gradients max_norm (float or int): max norm of the gradients
norm_type (float or int): type of the used p-norm. Can be ``'inf'`` for norm_type (float or int): type of the used p-norm. Can be ``'inf'`` for
infinity norm. infinity norm.
model_parallel_group (group): due to the nature of the distributed model_parallel_group (group): given the nature of the distributed
optimizer, this is passed as an argument. optimizer, this is passed as an argument.
Returns: Returns:
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment