| # 1. Start from the same great NVIDIA base image | |
| FROM nvcr.io/nvidia/pytorch:24.04-py3 | |
| # 2. Set the working directory | |
| WORKDIR /repository | |
| # 3. Add Hugging Face specific environment variables | |
| ENV PROTOCOL_BUFFERS_PYTHON_IMPLEMENTATION=python | |
| ENV PORT=80 | |
| # 4. Copy and install requirements AND the HF toolkit | |
| COPY requirements.txt . | |
| RUN pip install --no-cache-dir -r requirements.txt huggingface_inference_toolkit | |
| # 5. Copy the rest of your application code | |
| COPY . . | |
| # 6. THE CRUCIAL PART: Set the command to run the inference server | |
| CMD ["python", "-m", "huggingface_inference_toolkit.main", "--model_dir", "."] |