Fetching the paper…

Efficient Heterogeneous Large Language Model Decoding with Model-Attention Disaggregation · Around