Fix(docker): start of the model when building the Docker image

Dockerfile

@@ -5,12 +5,13 @@ RUN useradd -m -u 1000 user
 USER user
 ENV PATH="/home/user/.local/bin:$PATH"
 
-# Install system dependencies needed for many ML/AI packages
+# Install system dependencies
 USER root
 RUN apt-get update && apt-get install -y \
     build-essential \
     gcc \
     g++ \
+    cmake \
     libffi-dev \
     libsm6 \
     libxext6 \
@@ -31,11 +32,20 @@ WORKDIR /app
 # Copy and install dependencies
 COPY --chown=user ./requirements.txt requirements.txt
 RUN pip install --upgrade pip \
+    && CMAKE_ARGS="-DLLAMA_CUBLAS=off" pip install llama-cpp-python \
     && pip install --no-cache-dir -r requirements.txt
 
+# Copy models download script
+COPY --chown=user ./models.sh models.sh
+USER root
+RUN chmod +x /app/models.sh
+USER user
+
+# Download models
+RUN ./models.sh
+
 # Copy the rest of the project
 COPY --chown=user . /app
 
-# Default command to run the app
-# Notice we point to "app.main:app" because main.py is inside the "app" folder
-CMD ["gunicorn", "app.main:app", "-k", "uvicorn.workers.UvicornWorker", "--bind", "0.0.0.0:7860", "--workers", "2"]
+# Default command
+CMD ["gunicorn", "app.main:app", "-k", "uvicorn.workers.UvicornWorker", "--bind", "0.0.0.0:7860", "--workers", "2"]