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

moe args fix #54

Merged
merged 1 commit into from
Feb 3, 2025
Merged

moe args fix #54

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
5 changes: 3 additions & 2 deletions megatron/training/arguments.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2348,8 +2348,6 @@ def get_patch_args(parser):
),
)

group.add_argument("--moe-ffn-hidden-size", type=int, default=None)

group.add_argument("--shared-moe-ffn-hidden-size", type=int, default=None)

group.add_argument(
Expand Down Expand Up @@ -2568,6 +2566,9 @@ def _add_moe_args(parser):
'where 1 indicates an expert layer and 0 indicates a dense layer. '
'Examples: "([0]+[1]*23)": 1 dense layer followed by 23 experts layers, '
'"([1]*3+[0]*2)*2": Three expert layers followed by two dense layers, repeated twice.')
group.add_argument('--moe-ffn-hidden-size', type=int, default=None,
help='The hidden size of each expert\'s feed-forward network (ffn). '
'If not specified, defaults to the ffn_hidden_size.')
group.add_argument('--moe-shared-expert-intermediate-size', type=int, default=None,
help='Shared expert total ffn hidden size. '
'It should be equal to "num_shared_experts * ffn_size_of_each_shared_expert" if there are multiple shared experts. '
Expand Down