Getting dtype mismatch error when trying to use fp8_unet

[bradsevers]

I'm new to this so I may have something set up incorrectly. But when I try to run SUPIR using the "fp8_unet" setting, I get this error:

Error occurred when executing SUPIR_sample:

mat1 and mat2 must have the same dtype, but got Float and Float8_e4m3fn

File "C:\ComfyUI\ComfyUI_windows_portable\ComfyUI\execution.py", line 151, in recursive_execute
output_data, output_ui = get_output_data(obj, input_data_all)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\ComfyUI\ComfyUI_windows_portable\ComfyUI\execution.py", line 81, in get_output_data
return_values = map_node_over_list(obj, input_data_all, obj.FUNCTION, allow_interrupt=True)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\ComfyUI\ComfyUI_windows_portable\ComfyUI\execution.py", line 74, in map_node_over_list
results.append(getattr(obj, func)(**slice_dict(input_data_all, i)))
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\ComfyUI\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-SUPIR\nodes_v2.py", line 481, in sample
_samples = self.sampler(denoiser, noised_z, cond=positive[i], uc=negative[i], x_center=sample.unsqueeze(0), control_scale=control_scale_end,
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\ComfyUI\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-SUPIR\sgm\modules\diffusionmodules\sampling.py", line 441, in __call__
x = self.sampler_step(
^^^^^^^^^^^^^^^^^^
File "C:\ComfyUI\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-SUPIR\sgm\modules\diffusionmodules\sampling.py", line 418, in sampler_step
denoised = self.denoise(x, denoiser, sigma_hat, cond, uc, control_scale=control_scale)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\ComfyUI\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-SUPIR\sgm\modules\diffusionmodules\sampling.py", line 400, in denoise
denoised = denoiser(*self.guider.prepare_inputs(x, sigma, cond, uc), control_scale)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\ComfyUI\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-SUPIR\nodes_v2.py", line 455, in
denoiser = lambda input, sigma, c, control_scale: SUPIR_model.denoiser(SUPIR_model.model, input, sigma, c, control_scale)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\ComfyUI\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-SUPIR\sgm\modules\diffusionmodules\denoiser.py", line 73, in __call__
return network(input * c_in, c_noise, cond, control_scale) * c_out + input * c_skip
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\ComfyUI\ComfyUI_windows_portable\python_embeded\Lib\site-packages\torch\nn\modules\module.py", line 1518, in _wrapped_call_impl
return self._call_impl(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\ComfyUI\ComfyUI_windows_portable\python_embeded\Lib\site-packages\torch\nn\modules\module.py", line 1527, in _call_impl
return forward_call(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\ComfyUI\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-SUPIR\sgm\modules\diffusionmodules\wrappers.py", line 91, in forward
control = self.control_model(x=c.get("control", None), timesteps=t, xt=x,
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\ComfyUI\ComfyUI_windows_portable\python_embeded\Lib\site-packages\torch\nn\modules\module.py", line 1518, in _wrapped_call_impl
return self._call_impl(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\ComfyUI\ComfyUI_windows_portable\python_embeded\Lib\site-packages\torch\nn\modules\module.py", line 1527, in _call_impl
return forward_call(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\ComfyUI\ComfyUI_windows_portable\ComfyUI\custom_nodes\ComfyUI-SUPIR\SUPIR\modules\SUPIR_v0.py", line 519, in forward
emb = self.time_embed(t_emb)
^^^^^^^^^^^^^^^^^^^^^^
File "C:\ComfyUI\ComfyUI_windows_portable\python_embeded\Lib\site-packages\torch\nn\modules\module.py", line 1518, in _wrapped_call_impl
return self._call_impl(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\ComfyUI\ComfyUI_windows_portable\python_embeded\Lib\site-packages\torch\nn\modules\module.py", line 1527, in _call_impl
return forward_call(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\ComfyUI\ComfyUI_windows_portable\python_embeded\Lib\site-packages\torch\nn\modules\container.py", line 215, in forward
input = module(input)
^^^^^^^^^^^^^
File "C:\ComfyUI\ComfyUI_windows_portable\python_embeded\Lib\site-packages\torch\nn\modules\module.py", line 1518, in _wrapped_call_impl
return self._call_impl(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\ComfyUI\ComfyUI_windows_portable\python_embeded\Lib\site-packages\torch\nn\modules\module.py", line 1527, in _call_impl
return forward_call(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\ComfyUI\ComfyUI_windows_portable\python_embeded\Lib\site-packages\torch\nn\modules\linear.py", line 114, in forward
return F.linear(input, self.weight, self.bias)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

Here is a PNG containing my workflow (it's basically the same as the example SDXL Lightning workflow provided in this repo):

[bradsevers]

I forgot to mention the included image is the result of the denoise stage.