To run an AI node locally, you need to have the Unchained client installed. You'll
also need to have installed the Unchained AI plugin and the additional Broker
dependencies.
Create a new directory for your Unchained AI testnet and navigate to it:
mkdir unchained-ai
cd unchained-ai
First, generate a certificate for the broker by running the following command:
mkcert -install
mkcert localhost
Then, create a configuration file for the broker named `broker.yaml` with the
following content:
system:
log: info
name: YOUR_BROKER_NAME
network:
bind: 127.0.0.1:9123
certFile: localhost.pem
keyFile: localhost-key.pem
rpc:
- name: arbitrumSepolia
nodes:
- https://sepolia-rollup.arbitrum.io/rpc
Finally, run the broker with the following command:
unchained broker -c broker.yaml -a
Navigate to the directory where you cloned the `py-unchained-ai` repository and run
the following command:
source venv/bin/activate
python src/main.py us --file /tmp/unchained.sock
Create a configuration file for the worker named `worker.yaml` with the following
content:
system:
log: info
name: YOUR_WORKER_NAME
network:
brokerUri: wss://localhost:9123
rpc:
- name: arbitrumSepolia
nodes:
- https://sepolia-rollup.arbitrum.io/rpc
functions:
- name: Unchained.AI.TextToImage
endpoint: /tmp/unchained.sock
type: unix
plugins:
uniswap:
schedule:
Finally, run the worker with the following command:
unchained worker -c worker.yaml
Head over to the Unchained GUI , connect your wallet, and start
using the AI features.
The following AI models are available for use with the Unchained AI plugin:
- segmind/SSD-1B
- Corcelio/mobius
- segmind/Segmind-Vega
- Corcelio/openvision
- SimianLuo/LCM_Dreamshaper_v7
- OEvortex/PixelGen
- fluently/Fluently-XL-Final
- alvdansen/littletinies
- cagliostrolab/animagine-xl-3.1
- SG161222/Realistic_Vision_V6.0_B1_noVAE
- Lykon/dreamshaper-xl-v2-turbo
- UnfilteredAI/NSFW-gen-v2.1
Please note that the availability of AI models may change over time. When using a
model for the first time, it may take a while to download the model files. You can
follow the progress in the logs of the AI plugin. Make sure to have enough disk space
available for the downloaded model files.