1580158
/
ComfyUI
mirror of https://github.com/comfyanonymous/ComfyUI
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

fix: use Parameter assignment for Stable_Zero123 cc_projection weights (fixes #13492) (#13518) 

On Windows with aimdo enabled, disable_weight_init.Linear uses lazy
initialization that sets weight and bias to None to avoid unnecessary
memory allocation. This caused a crash when copy_() was called on the
None weight attribute in Stable_Zero123.__init__.

Replace copy_() with direct torch.nn.Parameter assignment, which works
correctly on both Windows (aimdo enabled) and other platforms.
pull/13471/merge
Octopus 2026-04-23 06:05:43 +08:00 committed by GitHub
parent db85cf03ff
commit cc6f9500a1
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 2 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
comfy/model_base.py
View File

@ -578,8 +578,8 @@ class Stable_Zero123(BaseModel):
def __init__(self, model_config, model_type=ModelType.EPS, device=None, cc_projection_weight=None, cc_projection_bias=None):
super().__init__(model_config, model_type, device=device)
self.cc_projection = comfy.ops.manual_cast.Linear(cc_projection_weight.shape[1], cc_projection_weight.shape[0], dtype=self.get_dtype(), device=device)
self.cc_projection.weight.copy_(cc_projection_weight)
self.cc_projection.bias.copy_(cc_projection_bias)
self.cc_projection.weight = torch.nn.Parameter(cc_projection_weight.clone())
self.cc_projection.bias = torch.nn.Parameter(cc_projection_bias.clone())
def extra_conds(self, **kwargs):
out = {}