text-generation-inference

f08b44ad - Upgrade to new vllm extension ops for Gaudi backend (fix issue in exponential bucketing) (#3239)