Skip to content

revert of #2658 #3736

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
sbak5 merged 2 commits into from
Mar 9, 2026
Merged

revert of #2658 #3736

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
64d16d5
revert of 2658
dimapihtar Mar 6, 2026
cfe9bf6
Merge branch 'main' into revert_2658
dimapihtar Mar 6, 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
23 changes: 2 additions & 21 deletions megatron/training/checkpointing.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -360,27 +360,8 @@ def get_rng_state(ckpt_format: str, tp_group: torch.distributed.ProcessGroup, pp
pp_size = get_pg_size(pp_group)
tp_rank = get_pg_rank(tp_group)
tp_size = get_pg_size(tp_group)
ep_size = mpu.get_expert_model_parallel_world_size()

if ep_size > 1:
# Shard RNG by PP, TP, DP when using expert parallelism.
dp_rank = mpu.get_data_parallel_rank(with_context_parallel=True)
dp_size = mpu.get_data_parallel_world_size(with_context_parallel=True)
rng_state_list = ShardedObject(
'rng_state',
rng_state_list,
(pp_size, tp_size, dp_size),
(pp_rank, tp_rank, dp_rank),
replica_id=0,
)
else:
rng_state_list = ShardedObject(
'rng_state',
rng_state_list,
(pp_size, tp_size),
(pp_rank, tp_rank),
replica_id=mpu.get_data_parallel_rank(with_context_parallel=True),
)
rng_state_list = ShardedObject('rng_state', rng_state_list, (pp_size, tp_size), (pp_rank, tp_rank),
replica_id=mpu.get_data_parallel_rank(with_context_parallel=True))
elif ckpt_format == "fsdp_dtensor":
pp_rank = mpu.get_pipeline_model_parallel_rank()
tp_rank = mpu.get_tensor_model_parallel_rank()
Expand Down
Loading