Commit Graph

21 Commits

Author SHA1 Message Date
oobabooga
78c0da4a18
Use the cuda branch of gptq-for-llama
Did I do this right @jllllll? This is because the current default branch (triton) is not compatible with Windows.
2023-03-30 18:04:05 -03:00
jllllll
ed0e593161
Change Micromamba download
Changed link to previous version.
This will provide a stable source for Micromamba so that new versions don't cause issues.
2023-03-29 02:47:19 -05:00
jllllll
cb5dff0087
Update installer to use official micromamba url 2023-03-26 23:40:46 -05:00
jllllll
bdf85ffcf9
Remove explicit pytorch installation
Fixes an issue some people were having: https://github.com/oobabooga/text-generation-webui/issues/15
I did not experience this issue on my system. Not everyone does for some reason.
2023-03-26 21:56:16 -05:00
jllllll
6f89242094
Remove temporary fix for GPTQ-for-LLaMa
No longer necessary.
2023-03-26 03:29:14 -05:00
jllllll
6dcfcf4fed
Amended fix for GPTQ-for-LLaMa
Prevents breaking 3-bit support
2023-03-26 01:00:52 -05:00
jllllll
12baa0e84b
Update for latest GPTQ-for-LLaMa 2023-03-26 00:46:07 -05:00
jllllll
247e8e5b79
Fix for issue in current GPTQ-for-LLaMa. 2023-03-26 00:24:00 -05:00
jllllll
ce9a5e3b53
Update install.bat
Minor fixes
2023-03-25 02:22:02 -05:00
jllllll
2e02d42682 Changed things around to allow Micromamba to work with paths containing spaces. 2023-03-25 01:26:25 -05:00
jllllll
1e260544cd
Update install.bat
Added C:\Windows\System32 to PATH to avoid issues with broken? Windows installs.
2023-03-24 21:25:14 -05:00
jllllll
eec773b1f4
Update install.bat
Corrected libbitsandbytes_cudaall.dll install.
2023-03-24 17:54:47 -05:00
jllllll
817e6c681e
Update install.bat
Added `cd /D "%~dp0"` in case the script is ran as admin.
2023-03-24 17:51:13 -05:00
jllllll
a80a5465f2
Update install.bat
Updated Conda packages and channels to install cuda-toolkit and override 12.0 cuda packages requested by pytorch with their 11.7 equivalent.
Removed Conda installation since we can use the downloaded Micromamba.exe for the same purpose with a smaller footprint.
Removed redundant PATH changes.
Changed %gpuchoice% comparisons to be case-insensitive.
Added additional error handling and removed the use of .tmp files.
Added missing extension requirements.
Added GPTQ installation. Will attempt to compile locally and, if failed, will download and install a precompiled wheel.
Incorporated fixes from one-click-bandaid.
Fixed and expanded first sed command from one-click-bandaid.
libbitsandbytes_cudaall.dll is used here as the cuda116.dll used by one-click-bandaid does not work on my 1080ti. This can be changed if needed.
2023-03-24 17:27:29 -05:00
oobabooga
9ed3a03d4b
Don't use the official instructions 2023-03-18 11:25:08 -03:00
oobabooga
91371640f9
Use the official instructions
https://pytorch.org/get-started/locally/
2023-03-17 20:37:25 -03:00
oobabooga
4639bc84c0
Specify torchvision version 2023-03-15 20:08:35 -03:00
oobabooga
92d8b3e8dd
Update install.bat 2023-03-13 22:53:06 -03:00
oobabooga
3e71099402
Add --upgrade everywhere 2023-03-13 22:50:42 -03:00
oobabooga
88b2520fda
Add --upgrade to pip 2023-03-13 22:49:13 -03:00
oobabooga
88af917e0e
Add files via upload 2023-03-12 16:42:50 -03:00