Hierarchical Verification of Speculative Beams for Accelerating LLM Inference

Re-assign community