Build and push docker images to GHCR

.github/workflows/docker_build_push.yaml

@@ -0,0 +1,36 @@
+name: Build Docker Image and Push to GHCR
+
+on:
+  push:
+    branches:
+    - main
+    - feat/docker-build
+
+jobs:
+  build:
+    runs-on: ubuntu-latest
+
+    steps:
+    - name: Checkout code
+      uses: actions/checkout@v2
+
+    - name: Set up Docker Buildx
+      uses: docker/setup-buildx-action@v1
+
+    - name: Login to GitHub Container Registry
+      uses: docker/login-action@v2
+      with:
+        registry: ghcr.io
+        username: ${{ github.actor }}
+        password: ${{ secrets.GITHUB_TOKEN }}
+
+    - name: Build and push Docker image
+      uses: docker/build-push-action@v4
+      with:
+        context: .
+        push: true
+        tags: |
+          ghcr.io/developmentseed/llllm:latest
+          ghcr.io/developmentseed/llllm:${{ github.sha }}
+        cache-from: type=gha
+        cache-to: type=gha,mode=max

Dockerfile

@@ -0,0 +1,20 @@
+FROM mambaorg/micromamba:1.4-bullseye-slim
+
+USER root
+RUN apt update && apt install -y gcc python3-dev \
+    && rm -rf /var/lib/apt/lists/*
+USER $MAMBA_USER
+
+COPY --chown=$MAMBA_USER:$MAMBA_USER environment.yaml /tmp/env.yaml
+# TODO: avoid installing development dependencies
+RUN micromamba env create --yes -f /tmp/env.yaml && \
+    micromamba clean --all --yes
+
+EXPOSE 8501
+
+COPY . /app
+WORKDIR /app
+
+HEALTHCHECK CMD curl --fail http://localhost:8501/_stcore/health
+
+ENTRYPOINT ["/opt/conda/envs/llllm-env/bin/streamlit", "run", "app.py", "--server.port=8501", "--server.address=0.0.0.0"]

agents/l4m_agent.py

@@ -1,7 +1,8 @@
 from langchain.agents import initialize_agent
+from langchain.agents import AgentType
 
 
-def base_agent(llm, tools, name="zero-shot-react-description"):
+def base_agent(llm, tools, agent_type=AgentType.STRUCTURED_CHAT_ZERO_SHOT_REACT_DESCRIPTION):
     """Base agent to perform xyz slippy map tiles operations.
 
     llm: LLM object
@@ -10,7 +11,7 @@ def base_agent(llm, tools, name="zero-shot-react-description"):
     agent = initialize_agent(
         llm=llm,
         tools=tools,
-        agent=name,
+        agent=agent_type,
         max_iterations=3,
         early_stopping_method="generate",
         verbose=True,

app.py

@@ -1,8 +1,10 @@
 import os
 
-import osmnx as ox
-import geopandas as gpd
+import folium
+import streamlit as st
+from streamlit_folium import folium_static
 
+from langchain.agents import AgentType
 from langchain.chat_models import ChatOpenAI
 from langchain.tools import Tool, DuckDuckGoSearchRun
 
@@ -13,11 +15,6 @@
 from tools.osmnx.network import OSMnxNetworkTool
 from agents.l4m_agent import base_agent
 
-import geopandas as gpd
-import streamlit as st
-import folium
-from streamlit_folium import folium_static
-
 
 def get_llm():
     llm = ChatOpenAI(
@@ -28,7 +25,7 @@ def get_llm():
     return llm
 
 
-def get_agent(llm, name="structured-chat-zero-shot-react-description"):
+def get_agent(llm, agent_type=AgentType.STRUCTURED_CHAT_ZERO_SHOT_REACT_DESCRIPTION):
     # define a set of tools the agent has access to for queries
     duckduckgo_tool = Tool(
         name="DuckDuckGo",
@@ -51,7 +48,7 @@ def get_agent(llm, name="structured-chat-zero-shot-react-description"):
         network_tool,
     ]
 
-    agent = base_agent(llm, tools, name=name)
+    agent = base_agent(llm, tools, agent_type=agent_type)
     return agent
 
 

environment.yaml

@@ -6,12 +6,13 @@ dependencies:
   - pip
   - osmnx=1.3.1
   - pip:
-    - openai
-    - langchain
-    - duckduckgo-search
-    - mercantile
-    - geopy
-    - ipywidgets
-    - jupyterlab
-    - streamlit
-    - streamlit-folium
+    - openai==0.27.8
+    - langchain==0.0.215
+    - duckduckgo-search==3.8.3
+    - mercantile==1.2.1
+    - geopy==2.3.0
+    - ipywidgets==8.0.6
+    - jupyterlab==4.0.2
+    - streamlit==1.23.1
+    - streamlit-folium==0.12.0
+    - watchdog==3.0.0