Skip to main content

Academic Resource Acceleration

Public Services​

Github Acceleration Site: https://ghproxy.link

HuggingFace Acceleration Site: https://hf-mirror.com

GPUGEEK Academic Resource Acceleration​

Statement

GPUGEEK academic resource acceleration is limited to academic use for addressing slow network access to GitHub and HuggingFace. The following acceleration proxies are provided to facilitate academic use of related resources. There is no guarantee of stability. In case of malicious attacks, the acceleration service may be stopped at any time. This service is limited to use within GPUGEEK platform instances only.

The following academic resource domains (and their subdomains) can be accessed with acceleration:

google.com
huggingface.co
hf.co
nvidia.com
docker.io
docker.com
r8.im
replicate.com
python.org
nvcr.io
gcr.io
quay.io
k8s.io
ghcr.io
github.com
githubusercontent.com
pypi.org
pytorch.org
google-analytics.com
github.io
amazonaws.com
anaconda.com
anaconda.org
vllm.ai
microsoft.com
wandb.ai
gradio.app
azure.cn
googleapis.com
r8.im
us-docker.pkg.dev

Usage in Northwest Zone 1 (terminal):

curl -sSL -o /etc/network_turbo https://oss-high-qy01.cdsgss.com/public-static-prod-gpuzoom/installation/platform/cn-northwest1-network_turbo
chmod +x /etc/network_turbo
source /etc/network_turbo

Usage in East China Zone 3 (terminal):

curl -sSL -o /etc/network_turbo https://oss-high-qy01.cdsgss.com/public-static-prod-gpuzoom/installation/platform/cn-east3-network_turbo
chmod +x /etc/network_turbo
source /etc/network_turbo

Cancel Network Acceleration

unset http_proxy && unset https_proxy && unset all_proxy && unset no_proxy
unset HTTP_PROXY && unset HTTPS_PROXY && unset ALL_PROXY && unset NO_PROXY
git config --global --unset http.proxy
git config --global --unset https.proxy

Using Acceleration When Starting Programs​

For example, starting ComfyUI:

# Northwest Zone 1
source /etc/network_turbo && python3 main.py --listen 0.0.0.0 --port 8080

# East China Zone 3
source /etc/network_turbo && python3 main.py --listen 0.0.0.0 --port 8080