Leveraging Compute-in-Memory for Efficient Generative Model Inference in TPUs

Re-assign community