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

Fix bug causing random initialization of bias when using GPTQ quantization with models without bias #1827

Merged
merged 2 commits into from
Apr 29, 2024
Merged

Fix bug causing random initialization of bias when using GPTQ quantization with models without bias #1827

Changes from 1 commit
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
fe96b7c
Fix gptq quantization for models without bias
B-201 Apr 24, 2024
d573804
Fix gptq quantization for models without bias
B-201 Apr 28, 2024
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
5 changes: 3 additions & 2 deletions optimum/gptq/quantizer.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -278,19 +278,20 @@ def _replace_by_quant_layers(self, module: nn.Module, names: List[str], name: st
elif isinstance(layer, Conv1D):
in_features = layer.weight.shape[0]
out_features = layer.weight.shape[1]
bias = True if layer.bias else False
if not (self.desc_act) or self.group_size == -1:
new_layer = QuantLinear(
self.bits,
self.group_size,
in_features,
out_features,
True,
bias,
use_cuda_fp16=self.use_cuda_fp16,
weight_dtype=layer.weight.dtype,
)
else:
new_layer = QuantLinear(
self.bits, self.group_size, in_features, out_features, True, weight_dtype=layer.weight.dtype
self.bits, self.group_size, in_features, out_features, bias, weight_dtype=layer.weight.dtype
)
new_layer.device = device
setattr(module, attr, new_layer.to(device))
Expand Down
Loading