Fetching the paper…

SkipDecode: Autoregressive Skip Decoding with Batching and Caching for Efficient LLM Inference · Around