SPIRe: Boosting LLM Inference Throughput with Speculative Decoding

LRM
Re-assign community