Skip to content

[RL]Fix missing is_distributed attribute #5150

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
merged 2 commits into from
Nov 21, 2025
Merged

[RL]Fix missing is_distributed attribute #5150

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
3a31c57
fix
bukejiyu Nov 20, 2025
d745a2d
update
bukejiyu Nov 21, 2025
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
12 changes: 6 additions & 6 deletions fastdeploy/model_executor/layers/linear.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -56,9 +56,6 @@ def create_weights(self, layer: nn.Layer, **extra_weight_attrs):
is_bias=False,
default_initializer=paddle.nn.initializer.Constant(0),
)
split_axis = extra_weight_attrs.get("split_axis")
if hasattr(layer, "nranks") and layer.nranks > 0:
_set_var_distributed(layer.weight, split_axis=split_axis)

if self.model_format == "torch" and "output_dim" in extra_weight_attrs:
extra_weight_attrs["output_dim"] = not extra_weight_attrs["output_dim"]
Expand Down Expand Up @@ -882,15 +879,18 @@ def __init__(
if add_bias:
assert with_bias, "with_bias must be True when add_bias is True."
assert self.quant_method is not None
self.quant_method.create_weights(
self,
split_axis=0,
create_weight_kwargs = dict(
layer=self,
output_dim=None if self.split_token else False,
weight_loader=(
self.weight_loader if hasattr(self, "weight_loader") else default_weight_loader(self.fd_config)
),
model_format=fd_config.model_config.model_format,
)
if self.nranks > 0:
create_weight_kwargs["split_axis"] = 0
create_weight_kwargs["is_distributed"] = True
self.quant_method.create_weights(**create_weight_kwargs)

self.reduce_results = reduce_results

Expand Down
Loading