Skip to content

Support 4bit torch.compile fullgraph with PyTorch nightly #1616

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 1 commit into from
Apr 28, 2025
Merged

Support 4bit torch.compile fullgraph with PyTorch nightly #1616

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
9 changes: 8 additions & 1 deletion bitsandbytes/nn/modules.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -290,6 +290,13 @@ def from_prequantized(

return self

@classmethod
def __torch_function__(cls, func, types, args=(), kwargs=None):
if kwargs is None:
kwargs = {}
with torch._C.DisableTorchFunctionSubclass():
return func(*args, **kwargs)

def _quantize(self, device):
w = self.data.contiguous().to(device)
w_4bit, quant_state = bnb.functional.quantize_4bit(
Expand Down Expand Up @@ -486,7 +493,7 @@ def forward(self, x: torch.Tensor):

bias = None if self.bias is None else self.bias.to(self.compute_dtype)

return bnb.matmul_4bit(x, self.weight.t(), bias=bias, quant_state=self.weight.quant_state).to(inp_dtype)
return bnb.matmul_4bit(x, self.weight.data.t(), bias=bias, quant_state=self.weight.quant_state).to(inp_dtype)


class LinearFP4(Linear4bit):
Expand Down