Closed GamingDaveUk closed 1 year ago
Tried a fresh install, just before selecting a model I got the same error.
Noticed some people with similar issues needed c++ build tools. installed the full visual studio community edition made sure teh latest c++ build tools are installed in the individual modules section.
ran the updater.... same error
running the start bat and choosing a model shows this error:
`bin C:\AI\oobabooga_windows\installer_files\env\lib\site-packages\bitsandbytes\libbitsandbytes_cpu.dll C:\AI\oobabooga_windows\installer_files\env\lib\site-packages\bitsandbytes\cextension.py:33: UserWarning: The installed version of bitsandbytes was compiled without GPU support. 8-bit optimizers, 8-bit multiplication, and GPU quantization are unavailable. warn("The installed version of bitsandbytes was compiled without GPU support. " The following models are available:
Which one do you want to load? 1-6
5
INFO:Loading TheBloke_WizardLM-7B-uncensored-GPTQ...
CUDA extension not installed.
INFO:Found the following quantized model: models\TheBloke_WizardLM-7B-uncensored-GPTQ\WizardLM-7B-uncensored-GPTQ-4bit-128g.compat.no-act-order.safetensors
Traceback (most recent call last):
File "C:\AI\oobabooga_windows\text-generation-webui\server.py", line 984, in
Done! Press any key to continue . . .`
A couple of days ago this was working 100% I have no idea why it suddenly wont update or run, the fact a fresh install wont run either and that I seem to be the only one with this exact issue, makes me think its something on my system but no idea what
I'd give this a shot: https://github.com/oobabooga/text-generation-webui/issues/1289#issuecomment-1518719987
I'd give this a shot: #1289 (comment)
sadly this fails at the first hurdle
Microsoft Windows [Version 10.0.22621.1702]
(c) Microsoft Corporation. All rights reserved.
C:\AI\oobabooga_windows>python -m torch.utils.collect_env
Collecting environment information...
Traceback (most recent call last):
File "C:\Program Files\Python310\lib\runpy.py", line 196, in _run_module_as_main
return _run_code(code, main_globals, None,
File "C:\Program Files\Python310\lib\runpy.py", line 86, in _run_code
exec(code, run_globals)
File "C:\Users\mrgre\AppData\Roaming\Python\Python310\site-packages\torch\utils\collect_env.py", line 602, in <module>
main()
File "C:\Users\mrgre\AppData\Roaming\Python\Python310\site-packages\torch\utils\collect_env.py", line 585, in main
output = get_pretty_env_info()
File "C:\Users\mrgre\AppData\Roaming\Python\Python310\site-packages\torch\utils\collect_env.py", line 580, in get_pretty_env_info
return pretty_str(get_env_info())
File "C:\Users\mrgre\AppData\Roaming\Python\Python310\site-packages\torch\utils\collect_env.py", line 422, in get_env_info
pip_version, pip_list_output = get_pip_packages(run_lambda)
File "C:\Users\mrgre\AppData\Roaming\Python\Python310\site-packages\torch\utils\collect_env.py", line 394, in get_pip_packages
out = run_with_pip(sys.executable + ' -mpip')
File "C:\Users\mrgre\AppData\Roaming\Python\Python310\site-packages\torch\utils\collect_env.py", line 382, in run_with_pip
for line in out.splitlines()
AttributeError: 'NoneType' object has no attribute 'splitlines'
C:\AI\oobabooga_windows>
comfyui, automatic1111 all work fine so I assume torch and cuda is installed fine...but not sure
Correct me if I'm wrong, but from the log output you provided, it doesn't look like you used cmd_windows.bat.
But first, I would suggest you:
env
folder in: .\oobaboogaAI\installer_files\
start_windows.bat
to rebuild the environment. (You can then remove the old env folder if you renamed it).Correct me if I'm wrong, but from the log output you provided, it doesn't look like you used cmd_windows.bat.
But first, I would suggest you:
- Remove or rename the
env
folder in:.\oobaboogaAI\installer_files\
- Then click the
start_windows.bat
to rebuild the environment. (You can then remove the old env folder if you renamed it).- Once that is done, redo those commands using that bat file. Cuda extension not installed #1289 (comment)
I did not know about that bat file, usually i will just do cmd in the address bar and go from there. followed the steps you linked to in your earlier post (i have tried the fresh install and the deleting of the installer files before, tried tons of things in the last couple of days lol) The steps you linked too earlier allowed me to do the update with out error! yay and it loads too! Thankyou! I was starting to lose hope that it could be fixed with out a full system re-install
Ah it's easy to miss the conda script, just be sure to have the latest one-click installers, my previous environment got messed up and I was using the old installers, so a fresh install with the new scripts worked.
But glad to hear it's working, be sure to close this issue :)
The problem with updating is that after downloading GPTQ-for-LLaMa
repository, the file setup_cuda.py
gets renamed to setup.py
.
Next time you run update_windows.bat
, the git pull
command fails for that repository due to unstaged changes caused by this rename:
TL;DR — someone should submit a pull request for that repo to change the file name so it doesn't have to be done locally, or just change whatever logic is used to use the original file name without renaming it.
This issue has been closed due to inactivity for 30 days. If you believe it is still relevant, please leave a comment below.
well the issue remains, there is no easy to follow guide on getting qlora training or gptq 4bit training to work in textgen webui. There are the occasional posts that asay things like grab x and build its kernal... which makes little sense to people like myself (and I am an old geek)
I feel lora training is just abandoned at this stage.
Describe the bug
Was working fine for the last few days, and I update before each launch fine. today though i get an error updating. I am using the one click install version and updating via the bat file I did try to re-install the requirements.txt manually to see if that resolved it but it did not. The app has now lost the ability to load models.
Found a similar issue in the search that turned out the person was using the cpu version of pytorch, unless that has changed on my syatem (cant see why it would have) then i do not feel this is the same issue
Is there an existing issue for this?
Reproduction
run update_windows.bat
Screenshot
No response
Logs
System Info