mirrors/stable-diffusion-webui
1
Fork 0
mirror of https://github.com/AUTOMATIC1111/stable-diffusion-webui.git synced 2025-12-29 13:22:57 -08:00
Code Issues Projects Releases Packages Wiki Activity

Add FP32 fallback support on sd_vae_approx

Browse source 
This tries to execute interpolate with FP32 if it failed.

Background is that
on some environment such as Mx chip MacOS devices, we get error as follows:

```
"torch/nn/functional.py", line 3931, in interpolate
        return torch._C._nn.upsample_nearest2d(input, output_size, scale_factors)
               ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
    RuntimeError: "upsample_nearest2d_channels_last" not implemented for 'Half'
```

In this case, ```--no-half``` doesn't help to solve. Therefore this commits add the FP32 fallback execution to solve it.

Note that the submodule may require additional modifications. The following is the example modification on the other submodule.

```repositories/stable-diffusion-stability-ai/ldm/modules/diffusionmodules/openaimodel.py

class Upsample(nn.Module):
..snip..
    def forward(self, x):
        assert x.shape[1] == self.channels
        if self.dims == 3:
            x = F.interpolate(
                x, (x.shape[2], x.shape[3] * 2, x.shape[4] * 2), mode="nearest"
            )
        else:
            try:
                x = F.interpolate(x, scale_factor=2, mode="nearest")
            except:
                x = F.interpolate(x.to(th.float32), scale_factor=2, mode="nearest").to(x.dtype)
        if self.use_conv:
            x = self.conv(x)
        return x
..snip..
```

You can see the FP32 fallback execution as same as sd_vae_approx.py.
This commit is contained in:
hidenorly 2023-11-21 01:13:53 +09:00
parent 5f36f6ab21
commit 58c19545c8
1 changed files with 7 additions and 1 deletions
Download patch file Download diff file Expand all files Collapse all files

8
modules/sd_vae_approx.py
View file

@ -21,7 +21,13 @@ class VAEApprox(nn.Module):
def forward(self, x):
extra = 11
x = nn.functional.interpolate(x, (x.shape[2] * 2, x.shape[3] * 2))
try:
x = nn.functional.interpolate(x, (x.shape[2] * 2, x.shape[3] * 2))
except RuntimeError as e:
if "not implemented for" in str(e) and "Half" in str(e):
x = nn.functional.interpolate(x.to(torch.float32), (x.shape[2] * 2, x.shape[3] * 2)).to(x.dtype)
else:
print(f"An unexpected RuntimeError occurred: {str(e)}")
x = nn.functional.pad(x, (extra, extra, extra, extra))
for layer in [self.conv1, self.conv2, self.conv3, self.conv4, self.conv5, self.conv6, self.conv7, self.conv8, ]: