Awesome
Overview
This repo is a docker container for running SpikeGPT by ridgerchu with an Nvidia GPU. It's base is the Nvida PyTorch container which runs Ubuntu.
Prerequisites
Install Nvidia Container Toolkit for GPU Use wtih Docker
sudo apt-get update
sudo apt-get install -y nvidia-container-toolkit
sudo nvidia-ctk runtime configure --runtime=docker
sudo systemctl restart docker
Note: Installing the Nvidia Container Toolkit does not install the appropriate Nvidia drivers on your host machine for you. Be sure you can run nvidia-smi on the host machine with no errors.
Test the installation
sudo docker run --rm --runtime=nvidia --gpus all nvidia/cuda:11.6.2-base-ubuntu20.04 nvidia-smi
Your output should be similar to the following...
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 450.51.06 Driver Version: 450.51.06 CUDA Version: 11.0 |
|-------------------------------+----------------------+----------------------+
| GPU Name Persistence-M| Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. |
| | | MIG M. |
|===============================+======================+======================|
| 0 Tesla T4 On | 00000000:00:1E.0 Off | 0 |
| N/A 34C P8 9W / 70W | 0MiB / 15109MiB | 0% Default |
| | | N/A |
+-------------------------------+----------------------+----------------------+
+-----------------------------------------------------------------------------+
| Processes: |
| GPU GI CI PID Type Process name GPU Memory |
| ID ID Usage |
|=============================================================================|
| No running processes found |
+-----------------------------------------------------------------------------+
Running SpikeGPT Container
Copy the SpikeGPT directory
Be sure to include the approporate model weights as described in the SpikeGPT project README and copy the folder into the SpikeGPT-container directory
cd SpikeGPT-container
cp -R SpikeGPT .
Build and run the container
docker build -t spikegpt .
./run.bash
Once in the container simply launch the run script
python run.py