Dockerfile

# Use NVIDIA CUDA base image
FROM nvidia/cuda:12.1.0-runtime-ubuntu22.04 as base

# Set working directory to /code (Hugging Face Spaces convention)
WORKDIR /code

# Install system dependencies
RUN apt-get update && apt-get install -y \
    python3.10 \
    python3-pip \
    git \
    && rm -rf /var/lib/apt/lists/*

# Install Python packages
COPY requirements.txt .
RUN pip3 install --no-cache-dir -r requirements.txt

# Install any additional dependencies needed for litgpt
RUN pip3 install --no-cache-dir \
    einops \
    xformers \
    bitsandbytes \
    accelerate \
    sentencepiece

# Copy the application code
COPY . .

# Create model directory structure
RUN mkdir -p /code/checkout/meta \
    /code/checkout/microsoft \
    /code/checkout/mistralai

# Set environment variables
ENV PYTHONPATH=/code
ENV LLM_ENGINE_HOST=0.0.0.0
ENV LLM_ENGINE_PORT=7860

# Expose the port the app runs on
EXPOSE 8001

# Command to run the application
CMD ["python3", "main.py"]