Skip to content

[Bugfix] Fix Cosmos3 FP8 tensor parallelism #14921

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

Open
yzhautouskay wants to merge 1 commit into
base: main
Choose a base branch
from
Open

[Bugfix] Fix Cosmos3 FP8 tensor parallelism #14921

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
10 changes: 9 additions & 1 deletion examples/cosmos3/cosmos_parallel.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -431,7 +431,15 @@ def enable_cosmos3_tensor_parallel(transformer, tp_mesh):
own processor) on a 2-D ``(tp, cp)`` mesh, or with ``enable_cosmos3_flash_attention``
for TP without CP.
"""
from torch.distributed.tensor.parallel import ColwiseParallel, RowwiseParallel, parallelize_module
from torch.distributed.tensor.parallel import ColwiseParallel, parallelize_module
from torch.distributed.tensor.parallel import RowwiseParallel as TorchRowwiseParallel

# Skip ModelOpt quantizer children to avoid:
# AttributeError: 'TensorQuantizer' object has no attribute 'weight'
class RowwiseParallel(TorchRowwiseParallel):
def _partition_linear_fn(self, name, module, device_mesh):
if isinstance(module, torch.nn.Linear):
return super()._partition_linear_fn(name, module, device_mesh)

tp = tp_mesh.size()
dev = torch.device("cuda", torch.cuda.current_device())
Expand Down
Loading