LLama / Dockerfile
Hivra's picture
Create Dockerfile
4852b30 verified
raw
history blame contribute delete
509 Bytes
# Use a small Python image
FROM python:3.11-slim
# Set working folder
WORKDIR /app
# Copy only deps first for better caching
COPY requirements.txt .
# Install FastAPI, uvicorn, meta-ai-api, openai
RUN pip install --no-cache-dir -r requirements.txt
# Copy rest of code
COPY . .
# Let FastAPI know which model to use
# You can override this when you run the container
ENV MODEL_NAME=llama-3-70b
# Expose port
EXPOSE 7860
# Start the app
CMD ["uvicorn", "main:app", "--host", "0.0.0.0", "--port", "7860"]