Update README.md

docs: add model card with eval metrics on held-out test set

upload trained Qwen LangID guard LoRA

initial commit