Skip to content
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

Jump to bottom

[Bugfix] Fix a small device bug #57

Merged
merged 6 commits into from
Feb 16, 2023
Merged

[Bugfix] Fix a small device bug #57

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
6 commits
Select commit Hold shift + click to select a range
28322ab
fix a small bug
szhengac Feb 14, 2023
890cd66
add model parallel tag
szhengac Feb 15, 2023
4079635
add model parallel tag
szhengac Feb 15, 2023
825bbc8
fix
szhengac Feb 15, 2023
2e8f0bc
reformat
szhengac Feb 15, 2023
ffaa2df
fix
szhengac Feb 15, 2023
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
8 changes: 5 additions & 3 deletions slapo/framework_dialect/deepspeed/pipeline.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -316,9 +316,11 @@ def forward(self, *args, **kwargs):
if logger.isEnabledFor(DEBUG):
logger.debug(f"[{self.name}] Flatten: {len(ret)}; metadata: {metadata}")
ret.append(
torch.ByteTensor(
torch.ByteStorage.from_buffer(bytes(encode_metadata(metadata), "utf8"))
).to(device)
torch.tensor(
torch.ByteStorage.from_buffer(bytes(encode_metadata(metadata), "utf8")),
dtype=torch.uint8,
device=device,
)
)
ret = tuple(ret)
return ret
Expand Down
10 changes: 6 additions & 4 deletions slapo/schedule.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -261,10 +261,10 @@ def _shard(name, tensor):
self.metadata.tie_weights[param] = new_param
else:
new_param = nn.Parameter(new_tensor)

# Tag param with model parallel attribute, used for grad clipping
new_param.tensor_model_parallel = True
# Save the original size of the parameter for consolidation.
new_param.orig_shape = param.shape

self.mod.register_parameter(tensor_name, new_param)
except AttributeError:
buffer = self.mod.get_buffer(tensor_name)
Expand Down Expand Up @@ -1363,8 +1363,10 @@ def _consolidate_and_broadcast(sch: Schedule):
is_found = True
if is_found:
cnt_shard += 1
new_param = param.detach().split(sharded_size, dim=axis)[tp_rank]
sch.mod.register_parameter(param_name, nn.Parameter(new_param))
sharded_param = param.detach().split(sharded_size, dim=axis)[tp_rank]
new_param = nn.Parameter(sharded_param)
new_param.tensor_model_parallel = True
sch.mod.register_parameter(param_name, new_param)

for subsch in sch.child.values():
ret = _consolidate_and_broadcast(subsch)
Expand Down
2 changes: 1 addition & 1 deletion slapo/sharding/sync_ops.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -41,7 +41,7 @@ def all_gather_along_dim(inp, dim, world_size, group):
else:
# Fallback to all_gather. This may lead to suboptimal performance.
parts = [
torch.empty(inp.shape, dtype=inp.dtype).to(inp.device)
torch.empty(inp.shape, dtype=inp.dtype, device=inp.device)
for _ in range(world_size)
]
dist.all_gather(parts, inp, group=group)
Expand Down