Dockerfile.gpu

FROM nvidia/cuda:12.6.3-cudnn-runtime-ubuntu24.04

# Install 'uv', which we will use to install Python dependencies
COPY --from=ghcr.io/astral-sh/uv:latest /uv /uvx /bin/

ENV PYTHONDONTWRITEBYTECODE="1" \
    PYTHONUNBUFFERED="1" \
    DEBIAN_FRONTEND="noninteractive"

# Install Python and create venv
WORKDIR /app/
RUN uv python install 3.12
RUN uv venv
RUN uv pip install --no-cache-dir --upgrade pip setuptools wheel

# Download models during build instead of copying from local
COPY scripts/model_download.bash /tmp/model_download.bash
RUN . .venv/bin/activate && \
    uv pip install --no-cache-dir huggingface-hub && \
    bash /tmp/model_download.bash && \
    rm /tmp/model_download.bash

# Install CPU requirements
COPY requirements.cpu.txt ./
RUN uv pip install --no-cache-dir -r ./requirements.cpu.txt

# Install GPU PyTorch requirements
COPY requirements.torch.gpu.txt ./
RUN uv pip install --no-cache-dir -r ./requirements.torch.gpu.txt

COPY app ./app
COPY main.py ./

EXPOSE 8000

ENTRYPOINT ["python3", "-m", "uvicorn", "main:app", "--host", "0.0.0.0", "--port", "8000"]