D2O: Dynamic Discriminative Operations for Efficient Long-Context Inference of Large Language Models

Devs

D2O: Dynamic Discriminative Operations for Efficient Long-Context Inference of Large Language Models | Read Paper on Bytez