KV Cache is 1 Bit Per Channel: Efficient Large Language Model Inference with Coupled Quantization

Add code
May 07, 2024

Share this with someone who'll enjoy it:

View paper onarxiv icon

Share this with someone who'll enjoy it: