Closed PrashantSaikia closed 11 months ago
Did you login with huggingface hub?
No, I did not explicitly login to HF hub, because -
Can you try with the latest version to see if you can still run into these issues?
at least the tests were coverred for linux and macos
what's the update, wondering were you able to run starcoder on macOS?
we need to support CPU (esp GGUF and GGML for it to run on MacOS)
This is on our roadmap but currently not very high priority atm.
Describe the bug
In Mac OS, starcoder does not even load, probably because it has no Nvidia GPU.
In Windows, the main issue is the dependency on the
bitsandbytes
library. Since the makers of that library never made a version for Windows, we have to rely on some hacks and tricks to get it to install on Windows. Which is something I had done in the past in my global environment with great difficulty; but since OpenLLM recommends to install it on a new conda environment, I will need to start all the way back from installing CUDA, which itself is a big hassle on Windows. So I gave up.Next, I tried on Google Colab notebook, which runs on Linux. Here,
bitsandbytes
is installed; however, this is the error message I get when I try tostart starcoder
:Full stacktrace provided in the Logs section below.
To reproduce
bitsandbytes
is installed correctly by trying to import it without error:import bitsandbytes
!openllm start starcoder
Logs
Environment
System information (Optional)
No response