search

Ree1s / IDM

279 stars 20 forks source link

The pre-trained div2k model mismatches the model config #32

Open CXNing opened 3 months ago

CXNing commented 3 months ago

RuntimeError: Error(s) in loading state_dict for GaussianDiffusion:
size mismatch for denoise_fn.ups.3.imnet.0.weight: copying a param with shape torch.Size([256, 516]) from checkpoint, the shape in current model is torch.Size([256, 514]).
size mismatch for denoise_fn.ups.7.imnet.0.weight: copying a param with shape torch.Size([256, 516]) from checkpoint, the shape in current model is torch.Size([256, 514]).
size mismatch for denoise_fn.ups.11.imnet.0.weight: copying a param with shape torch.Size([256, 260]) from checkpoint, the shape in current model is torch.Size([256, 258]).
size mismatch for denoise_fn.ups.15.imnet.0.weight: copying a param with shape torch.Size([256, 132]) from checkpoint, the shape in current model is torch.Size([256, 130])

CXNing commented 3 months ago

Could you please provide the corresponding Unet code for the pre-trained div2k model?