Skip to content

[FSDP2] Simplify shard_placement_fn in test #146847

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

[FSDP2] Simplify shard_placement_fn in test #146847

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
18 changes: 2 additions & 16 deletions test/distributed/_composable/fsdp/test_fully_shard_training.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -292,15 +292,7 @@ def _test_train_parity_single_group(
ref_optim = torch.optim.Adam(ref_model.parameters(), lr=1e-2)

def _shard_placement_fn(param: nn.Parameter) -> Optional[Shard]:
largest_dim = -1
largest_dim_size = -1
for dim, dim_size in enumerate(param.shape):
if dim_size > largest_dim_size:
largest_dim = dim
largest_dim_size = dim_size
assert largest_dim >= 0, f"{param.shape}"
assert largest_dim < param.ndim, f"{largest_dim=} {param.shape}"
return Shard(largest_dim)
return Shard(param.shape.index(max(param.shape)))

shard_placement_fn = _shard_placement_fn if use_shard_placement_fn else None
fully_shard(model, shard_placement_fn=shard_placement_fn)
Expand Down Expand Up @@ -769,13 +761,7 @@ def test_train_parity_shard_placement_fn_shard_largest_dim(self):
ref_optim = torch.optim.AdamW(ref_model.parameters(), lr=1e-2)

def shard_placement_fn(param: nn.Parameter) -> Optional[Shard]:
largest_dim = -1
largest_dim_size = -1
for dim, dim_size in enumerate(param.shape):
if dim_size > largest_dim_size:
largest_dim = dim
largest_dim_size = dim_size
return Shard(largest_dim)
return Shard(param.shape.index(max(param.shape)))

for layer in model.layers:
fully_shard(layer, shard_placement_fn=shard_placement_fn)
Expand Down
Loading