Skip to content

forward fix embedding_lookup to include DTensor #2167

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

Closed
wants to merge 1 commit into from
Closed

forward fix embedding_lookup to include DTensor #2167

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
19 changes: 18 additions & 1 deletion torchrec/distributed/embedding_lookup.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -23,6 +23,7 @@
from torch import nn

from torch.autograd.function import FunctionCtx
from torch.distributed._tensor import DTensor
from torch.nn.modules.module import _IncompatibleKeys
from torchrec.distributed.batched_embedding_kernel import (
BaseBatchedEmbedding,
Expand Down Expand Up @@ -69,7 +70,7 @@ def fx_wrap_tensor_view2d(x: torch.Tensor, dim0: int, dim1: int) -> torch.Tensor

def _load_state_dict(
emb_modules: "nn.ModuleList",
state_dict: "OrderedDict[str, Union[torch.Tensor, ShardedTensor]]",
state_dict: "OrderedDict[str, Union[torch.Tensor, ShardedTensor, DTensor]]",
) -> Tuple[List[str], List[str]]:
missing_keys = []
unexpected_keys = list(state_dict.keys())
Expand All @@ -95,6 +96,22 @@ def _load_state_dict(
)

dst_local_shard.tensor.detach().copy_(src_local_shard.tensor)
elif isinstance(dst_param, DTensor):
assert isinstance(src_param, DTensor)
dst_param = dst_param.to_local()
src_param = src_param.to_local()
assert len(dst_param.local_chunks) == len( # pyre-ignore[16]
src_param.local_chunks
)
for dst_local_shard, src_local_shard in zip(
dst_param.to_local().local_shards(), # pyre-ignore[16]
src_param.to_local().local_shards(),
):
assert (
dst_local_shard.metadata.local_chunks
== src_local_shard.metadata.local_chunks
)
dst_local_shard.detach().copy_(src_local_shard)
else:
assert isinstance(src_param, torch.Tensor) and isinstance(
dst_param, torch.Tensor
Expand Down