Open xiboliyaxiangjiaojun opened 1 month ago
Please try to use a proxy if you are in China mainland :)
Please try to use a proxy if you are in China mainland :)
I open the proxy, delete the original virtual environment and tried again, but still failed.
Please make sure your proxy works for the domain mlc.ai
There is no tar.gz file in https://mlc.ai/wheels, so it should still be a network problem
Please make sure your proxy works for the domain mlc.ai
There is no tar.gz file in https://mlc.ai/wheels, so it should still be a network problem
Maybe I can directly download .whl from https://mlc.ai/wheels, but I can't find one on it for arm64 and linux...
Hi @xiboliyaxiangjiaojun, we don't yet have the prebuild wheels for jetson, so as of now you need to build mlc from source. We will try to enable and prebuild wheels, which will take some time though.
先下载对应的包,再python -m pip install xxx.whl
I ran into this problem as well. Building from source isn't working either (can't find tvm.so etc.).
Could you please evaluate the build / install process on arm (RK3588 etc.) boards? There seem to be some recent issues. The original guide at https://blog.mlc.ai/2024/04/20/GPU-Accelerated-LLM-on-Orange-Pi doesn't seem to work anymore.
❓ General Questions
Hi, I’m trying to install mlc-llm on my Jetson agx orin. Environment: Jetson agx orin、Ubuntu 20.04、CUDA 12.2 I ran the following commands:
conda create -n mlc python=3.11
conda activate mlc
conda install -c conda-forge git-lfs
conda install -c conda-forge libgcc-ng
and thenpython -m pip install --pre -U -f https://mlc.ai/wheels mlc-llm-nightly-cu122 mlc-ai-nightly-cu122
It seems that I have successfully installed them.Installing collected packages: mlc-llm-nightly-cu122, mlc-ai-nightly-cu122 Successfully installed mlc-ai-nightly-cu122-0.1 mlc-llm-nightly-cu122-0.1
But I failed to import mlc_llmpython -c "import mlc_llm; print(mlc_llm)"
Traceback (most recent call last): File "<string>", line 1, in <module> ModuleNotFoundError: No module named 'mlc_llm'
pip list