Slim Dockerfile - no llama-cpp needed, just huggingface_hub + transformers for detector 7d3b867 verified Rofati commited on May 18