Improve docker deployment configuration (#163)

api_tokens_sample.yml

@@ -0,0 +1,2 @@
+api_key: # Insert api key here
+admin_key: # Insert admin key here

docker/.dockerignore

@@ -2,3 +2,5 @@ models/
 loras/
 .ruff_cache/
 **/__pycache__/
+config.yml
+api_tokens.yml

docker/Dockerfile

@@ -1,12 +1,5 @@
 # Use an official CUDA runtime with Ubuntu as a parent image
-FROM nvidia/cuda:12.4.1-runtime-ubuntu22.04
-
-ARG GIT_REPO=https://github.com/theroyallab/tabbyAPI
-ARG DO_PULL=true
-ENV DO_PULL $DO_PULL
-
-# Set the working directory in the container
-WORKDIR /app
+FROM nvidia/cuda:12.1.0-runtime-ubuntu22.04
 
 # Install system dependencies
 RUN apt-get update && apt-get install -y --no-install-recommends \
@@ -15,20 +8,14 @@ RUN apt-get update && apt-get install -y --no-install-recommends \
     ca-certificates \
     python3.11 \
     python3-pip \
-    git \
     && rm -rf /var/lib/apt/lists/*
 
-# Update repo
-RUN if [ ${DO_PULL} ]; then \
-    git init && \
-    git remote add origin $GIT_REPO && \
-    git fetch origin && \
-    git pull origin main && \
-    echo "Pull finished"; fi
-
 # Upgrade pip
 RUN pip3 install --no-cache-dir --upgrade pip
 
+# Set the working directory in the container
+WORKDIR /app
+
 # Get requirements
 COPY pyproject.toml .
 
@@ -38,9 +25,6 @@ RUN pip3 install --no-cache-dir .[cu121]
 # Copy the current directory contents into the container
 COPY . .
 
-# Create a config.yml
-COPY config_sample.yml config.yml
-
 # Make port 5000 available to the world outside this container
 EXPOSE 5000
 

docker/docker-compose.yml

@@ -12,7 +12,9 @@ services:
       - NAME=TabbyAPI
       - NVIDIA_VISIBLE_DEVICES=all
     volumes:
-      - ./models:/app/models
+      # - /path/to/models:/app/models                       # Change me
+      # - /path/to/config.yml:/app/config.yml               # Change me
+      # - /path/to/api_tokens.yml:/app/api_tokens.yml       # Change me
     deploy:
       resources:
         reservations: