RuntimeError: "LayerNormKernelImpl" not implemented for 'Half' #1349
Comments
|
hi, |
nikaskn=my username My specs nikaskn@hdhnl:~$ neofetch |
|
vine boom |
|
In the |
|
Full precision dramatically drops performance. What have I done: Gotten from https://github.com/AUTOMATIC1111/stable-diffusion-webui/wiki/Install-and-Run-on-NVidia-GPUs After follow https://pytorch.org/get-started/locally/ I tested on Win11 and error was gone with good performance. full precision does not required |
thx,I tried then it works! |
|
same question,i cannot run it on my amd cpu |
|
RuntimeError: "LayerNormKernelImpl" not implemented for 'Half' --precision full --no-half |
|
The reason is likely to be no GPU in your computer. I also encountered this problem, maybe it's better to download a GPU |
Can you copy the whole string? I tried to insert it in different ways, but it doesn't work. Maybe I'm writing the wrong way |
|
mezotaken
added
asking-for-help-with-local-system-issues
|
so, I am on an AMD system, which I am aware is not the best for this, but I've been trying to get it to work regardless, however, even with all the suggested solutions(commlandline_args set to full precision and no-half, and some other stuff) I still get the same error RuntimeError: "LayerNormKernelImpl" not implemented for 'Half' does anyone have any other ideas on how to get it to work? |
|
nvm that, for some reason it was not reading it from the webui-user file, upon adding it to the launch.py file it works now |
|
(WSL2 on Windows 10) |
|
@Sphinxxxx Thank you, this solved my problem. |
|
The last Two Lines should look like this. set COMMANDLINE_ARGS= --skip-torch-cuda-test --precision full --no-half call webui.bat |
Thanks, @Sphinxxxx! It works for me, I'm using Ubuntu 22.04. |
Hey @Ps43ffrFro , just run with this command, it doesn't need to be added to the config file. use |
|
it on work ,mac 2020, thanks! |
|
"./webui.sh --skip-torch-cuda-test --precision full --no-half" Where exactly does this go? |
|
@millbank14 - It's what you type when you start the application:
|
|
Thank you, it Worked! Really appreciate!👍🏼 |
|
it works on my case.
|
确实要在命令行加参数，直接在脚本里加参数如果不懂代码原理，很难加对 |
|
Propose: close this issue? or should this be detected and automatically use --no-half ? |
|
您好，您的邮件我已收到我将尽快查收并回复
|
|
Really appreciate! I worked for my case, where there's mac m1. |
|
您好，您的邮件我已收到我将尽快查收并回复
|
can someone show me or post a video on how to do this? because I think I'm doing it wrong and I keep trying but it's not working |
Hi @vonnie21, try the below command when startup.
|
|
great that worked! thank you |
|
i use it on colab so how should i fix it? |
|
您好，您的邮件我已收到我将尽快查收并回复
|
|
Is it possible to do it without using cpu, but gpu? Cause it's a bit slow though it works |
|
您好，您的邮件我已收到我将尽快查收并回复
|
|
Thank you, this worked but ruined the performance. it generates more than 15 minutes :( "./webui.sh --skip-torch-cuda-test --precision full --no-half" |
and others
Describe the bug
When I try to run any prompt in the webui, it shows me this message: "RuntimeError: "LayerNormKernelImpl" not implemented for 'Half'"
To Reproduce
Steps to reproduce the behavior:
1.Run webui-user.bat.
2.Get webui link.
3.Type prompt in txt2img and click generate.
4.See error.
Expected behavior
Images are generated.
Screenshots
ERRORLOG:
txt2img: A big turtle.
D:\SD 2.0\stable-diffusion-webui\venv\lib\site-packages\torch\amp\autocast_mode.py:198: UserWarning: User provided device_type of 'cuda', but CUDA is not available. Disabling
warnings.warn('User provided device_type of 'cuda', but CUDA is not available. Disabling')
Error completing request
Arguments: ('A big turtle.', '', 'None', 'None', 20, 0, False, False, 1, 1, 7, -1.0, -1.0, 0, 0, 0, False, 512, 512, False, False, 0.7, 0, False, None, '', False, 1, '', 4, '', True, False) {}
Traceback (most recent call last):
File "D:\SD 2.0\stable-diffusion-webui\modules\ui.py", line 142, in f
res = list(func(*args, **kwargs))
File "D:\SD 2.0\stable-diffusion-webui\webui.py", line 60, in f
res = func(*args, **kwargs)
File "D:\SD 2.0\stable-diffusion-webui\modules\txt2img.py", line 41, in txt2img
processed = process_images(p)
File "D:\SD 2.0\stable-diffusion-webui\modules\processing.py", line 351, in process_images
uc = prompt_parser.get_learned_conditioning(len(prompts) * [p.negative_prompt], p.steps)
File "D:\SD 2.0\stable-diffusion-webui\modules\prompt_parser.py", line 104, in get_learned_conditioning
conds = shared.sd_model.get_learned_conditioning(texts)
File "D:\SD 2.0\stable-diffusion-webui\repositories\stable-diffusion\ldm\models\diffusion\ddpm.py", line 558, in get_learned_conditioning
c = self.cond_stage_model(c)
File "D:\SD 2.0\stable-diffusion-webui\venv\lib\site-packages\torch\nn\modules\module.py", line 1130, in _call_impl
return forward_call(*input, **kwargs)
File "D:\SD 2.0\stable-diffusion-webui\modules\sd_hijack.py", line 387, in forward
outputs = self.wrapped.transformer(input_ids=tokens)
File "D:\SD 2.0\stable-diffusion-webui\venv\lib\site-packages\torch\nn\modules\module.py", line 1130, in _call_impl
return forward_call(*input, **kwargs)
File "D:\SD 2.0\stable-diffusion-webui\venv\lib\site-packages\transformers\models\clip\modeling_clip.py", line 722, in forward
return self.text_model(
File "D:\SD 2.0\stable-diffusion-webui\venv\lib\site-packages\torch\nn\modules\module.py", line 1130, in _call_impl
return forward_call(*input, **kwargs)
File "D:\SD 2.0\stable-diffusion-webui\venv\lib\site-packages\transformers\models\clip\modeling_clip.py", line 643, in forward
encoder_outputs = self.encoder(
File "D:\SD 2.0\stable-diffusion-webui\venv\lib\site-packages\torch\nn\modules\module.py", line 1130, in _call_impl
return forward_call(*input, **kwargs)
File "D:\SD 2.0\stable-diffusion-webui\venv\lib\site-packages\transformers\models\clip\modeling_clip.py", line 574, in forward
layer_outputs = encoder_layer(
File "D:\SD 2.0\stable-diffusion-webui\venv\lib\site-packages\torch\nn\modules\module.py", line 1130, in _call_impl
return forward_call(*input, **kwargs)
File "D:\SD 2.0\stable-diffusion-webui\venv\lib\site-packages\transformers\models\clip\modeling_clip.py", line 316, in forward
hidden_states = self.layer_norm1(hidden_states)
File "D:\SD 2.0\stable-diffusion-webui\venv\lib\site-packages\torch\nn\modules\module.py", line 1130, in _call_impl
return forward_call(*input, **kwargs)
File "D:\SD 2.0\stable-diffusion-webui\venv\lib\site-packages\torch\nn\modules\normalization.py", line 189, in forward
return F.layer_norm(
File "D:\SD 2.0\stable-diffusion-webui\venv\lib\site-packages\torch\nn\functional.py", line 2503, in layer_norm
return torch.layer_norm(input, normalized_shape, weight, bias, eps, torch.backends.cudnn.enabled)
RuntimeError: "LayerNormKernelImpl" not implemented for 'Half'
The text was updated successfully, but these errors were encountered: