Skip to main content

Run LLM application using Streamlit

Enable streamlit for your namespace​

We use streamlit as the tool for doing data analysis and LLM app playground, and we can create a separate streamlit environment for each namespace for better isolation.

  1. To enable streamlit for your namespace, add the annotation kubeagi.k8s.com.cn/streamlit.installed: "true" to the namespace.
apiVersion: v1
kind: Namespace
metadata:
  annotations:
    # enable streamlit for this namespace
    # change the value to 'false' will delete all resources related to this streamlit environment
    kubeagi.k8s.com.cn/streamlit.installed: "true"
  creationTimestamp: "2023-12-08T02:36:26Z"

Then there will be deployment, service and ingress created for streamlit under this namespace.

Access streamlit​

  1. Check the ingress configuration of the streamlit using kubectl get ing streamlit-app -n <your-namespace> -o yaml

  2. It'll have a context path similar as /chats/<your-namespace>, /chats is the global context configured at kubeagi-system's arcadia-config configMap.

So you can visit streamlit using address like https://portal.172.40.20.125.nip.io/chats/<your-namespace>

Customize your streamlit app​

Based on the streamlit design, we can customize streamlit application easily using python script.

  1. Check your streamlit pod
# get the pod
kubectl get pods -n <your-namespace>
# copy python script to your streamlit pages directory
# you can find some samples under examples/app-ui-using-streamlit directory
kubectl cp your-demo-streamlit-app.py <pod-name>:/app/pages/
  1. Then access the application using https://portal.172.40.20.125.nip.io/chats/<your-namespace>/your-demo-streamlit-app