Expected all tensors to be on the same device, but found at least two devices, cpu and cuda:0!

Nobody has claimed this yet.

Get error while trying to do something in controlnet integrated

Help me please! :(

System info: Win 10, RTX 2070 super 8 gb

Using model: dreamshaperXL_v21TurboDPMSDE.safetensors ,

Arguments:@echo off

set PATH=D:\ffmpeg-7.1\bin;%PATH%

set PYTHON=

set GIT=

set VENV_DIR=

set COMMANDLINE_ARGS=--xformers --medvram --cuda-stream --cuda-malloc --always-offload-from-vram --opt-sdp-attention --theme dark

git pull

call webui.bat

https://imgur.com/a/00loYDf

*** Error running process_before_every_sampling: D:\newstablediff\webui\extensions-builtin\sd_forge_controlnet\scripts\controlnet.py
Traceback (most recent call last):
File "D:\newstablediff\webui\modules\scripts.py", line 892, in process_before_every_sampling
script.process_before_every_sampling(p, *script_args, **kwargs)
File "D:\newstablediff\webui\venv\lib\site-packages\torch\utils\_contextlib.py", line 115, in decorate_context
return func(*args, **kwargs)
File "D:\newstablediff\webui\extensions-builtin\sd_forge_controlnet\scripts\controlnet.py", line 561, in process_before_every_sampling
self.process_unit_before_every_sampling(p, unit, self.current_params[i], *args, **kwargs)
File "D:\newstablediff\webui\venv\lib\site-packages\torch\utils\_contextlib.py", line 115, in decorate_context
return func(*args, **kwargs)
File "D:\newstablediff\webui\extensions-builtin\sd_forge_controlnet\scripts\controlnet.py", line 507, in process_unit_before_every_sampling
params.model.process_before_every_sampling(p, cond, mask, *args, **kwargs)
File "D:\newstablediff\webui\extensions-builtin\sd_forge_ipadapter\scripts\forge_ipadapter.py", line 147, in process_before_every_sampling
unet = opIPAdapterApply(
File "D:\newstablediff\webui\extensions-builtin\sd_forge_ipadapter\lib_ipadapter\IPAdapterPlus.py", line 688, in apply_ipadapter
clip_embed = clip_vision.encode_image(image)
File "D:\newstablediff\webui\backend\patcher\clipvision.py", line 123, in encode_image
outputs = self.model(pixel_values=pixel_values, output_hidden_states=True)
File "D:\newstablediff\webui\venv\lib\site-packages\torch\nn\modules\module.py", line 1532, in _wrapped_call_impl
return self._call_impl(*args, **kwargs)
File "D:\newstablediff\webui\venv\lib\site-packages\torch\nn\modules\module.py", line 1541, in _call_impl
return forward_call(*args, **kwargs)
File "D:\newstablediff\webui\venv\lib\site-packages\transformers\models\clip\modeling_clip.py", line 1483, in forward
vision_outputs = self.vision_model(
File "D:\newstablediff\webui\venv\lib\site-packages\torch\nn\modules\module.py", line 1532, in _wrapped_call_impl
return self._call_impl(*args, **kwargs)
File "D:\newstablediff\webui\venv\lib\site-packages\torch\nn\modules\module.py", line 1541, in _call_impl
return forward_call(*args, **kwargs)
File "D:\newstablediff\webui\venv\lib\site-packages\transformers\models\clip\modeling_clip.py", line 1029, in forward
hidden_states = self.embeddings(pixel_values)
File "D:\newstablediff\webui\venv\lib\site-packages\torch\nn\modules\module.py", line 1532, in _wrapped_call_impl
return self._call_impl(*args, **kwargs)
File "D:\newstablediff\webui\venv\lib\site-packages\torch\nn\modules\module.py", line 1541, in _call_impl
return forward_call(*args, **kwargs)
File "D:\newstablediff\webui\venv\lib\site-packages\transformers\models\clip\modeling_clip.py", line 195, in forward
embeddings = torch.cat([class_embeds, patch_embeds], dim=1)
RuntimeError: Expected all tensors to be on the same device, but found at least two devices, cpu and cuda:0! (when checking argument for argument tensors in method wrapper_CUDA_cat)

Reproduce the ControlNet/IP-Adapter path described in extensions-builtin/sd_forge_controlnet/scripts/controlnet.py and extensions-builtin/sd_forge_ipadapter/lib_ipadapter/IPAdapterPlus.py. Start at clipvision.py:123 and the Transformers CLIP stack trace, then determine where CPU and cuda:0 tensors diverge; done means this sampling workflow no longer raises the device-mismatch error.

Written by the indexing model from the issue text.