Open bossrhino opened 8 months ago
you need visual studios
I have the same error and do have VS 2022 installed but still receive the error. I know in https://github.com/oobabooga/text-generation-webui it is solved programmatically with different system and decision trees. Maybe the coders of privateGPT can have a look into it? Thx in advance for supporting.
@sumitsodhi88 I have Visual Studios installed. Specifically Visual Studio Code. Unless I'm missing something, this is the only one available for Ubuntu.
Stale issue
Stale because no one has responded
For me, the solution was to include the "-ccbin clang-14' to nvcc like the error said, so doing the following on a Debian 12 system allowed me to compile llama-cpp-python with GPU support.
Just add NVCC_PREPEND_FLAGS='-ccbin clang-14' like so:
CMAKE_ARGS='-DLLAMA_CUBLAS=ON' NVCC_PREPEND_FLAGS='-ccbin clang-14' poetry run pip install --force-reinstall --no-cache-dir llama-cpp-python
OS: Ubuntu 23.10 GPU: Nvidia 3070
Nvidia NVCC version 12.0 Build: cuda_12.0.r12.0/compiler.32267302_0
nvidia-smi is working.
gcc version: 13.2.0 g++ version: 13.2.0
References:
Command to replicate problem:
Note: In the installation doc I'm on the 'Linux NVIDIA GPU support and Windows-WSL' section.
I'm working on my linux desktop to install private gpt. Followed most of the tutorials on the docs page + novaspirits youtube video. I am running this in a python virtual environment, so I'm not sure if that has something to do with this error. But this is the error I'm stuck on:
Appreciate any advice, i've not seen this issue in the issues thread yet.