Skip to content

compat megatron.core 0.18 #77

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
Jintao-Huang merged 2 commits into from
May 14, 2026
+3 −0
Merged

compat megatron.core 0.18 #77

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
869a141
compat megatron.core 0.18
Jintao-Huang May 14, 2026
f995923
fix
Jintao-Huang May 14, 2026
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: 3 additions & 0 deletions src/mcore_bridge/model/modules/transformer_layer.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2,6 +2,7 @@
import enum
import inspect
import torch
from functools import partial
from megatron.core.extensions.transformer_engine import TEFusedMLP
from megatron.core.process_groups_config import ProcessGroupCollection
from megatron.core.tensor_parallel.mappings import (gather_from_sequence_parallel_region,
Expand Down Expand Up @@ -210,6 +211,8 @@ def can_recompute_pre_mlp_layernorm_for_cudagraph():

def _build_mlp(self, mlp_spec):
pg_collection = self.pg_collection
if isinstance(mlp_spec, partial):
return mlp_spec(config=self.config, pg_collection=pg_collection, is_mtp_layer=self.is_mtp_layer)
additional_mlp_kwargs = {}
# import here to avoid circular import
from mcore_bridge.model.gpts.glm4 import Glm4MLP
Expand Down
Loading