Dockerfile

# Use an official Python runtime as the base image
FROM python:3.10

# Set environment variables for Hugging Face cache (FIX: Use /tmp)
ENV TRANSFORMERS_CACHE=/tmp
ENV HF_HOME=/tmp
ENV HF_MODULES_CACHE=/tmp

# Set the working directory in the container
WORKDIR /app

# Copy the FastAPI application code
COPY server.py /app/server.py

# Install necessary dependencies
RUN pip install --no-cache-dir fastapi uvicorn transformers torch sentencepiece

# Expose the FastAPI port
EXPOSE 7860

# Run the FastAPI server
CMD ["uvicorn", "server:app", "--host", "0.0.0.0", "--port", "7860"]