text-generation-inference

aa88c4fd - fix: add lora kernel to dockerfile, support running without kernels and refactors