FROM ghcr.io/huggingface/text-embeddings-inference:cpu-1.7

# Create data directory and set permissions
USER root
RUN mkdir -p /data && \
    chown -R 1000:1000 /data && \
    chmod 755 /data

ENV MODEL_ID=nomic-ai/nomic-embed-text-v1.5
ENV HF_HUB_CACHE=/data


# Switch back to non-root user
USER 1000

ENTRYPOINT ["text-embeddings-router"]
CMD ["--model-id", "nomic-ai/nomic-embed-text-v1.5", \
     "--port", "7860", \
     "--hostname", "0.0.0.0", \
     "--max-concurrent-requests", "512", \
     "--max-batch-tokens", "16384", \
     "--max-client-batch-size", "32", \
     "--huggingface-hub-cache", "/data"]