Skip to content

Accurate, fast FP4 KV-cache decoding through algorithm-system co-design.

Describe algorithm-system co-design in README

DD-DuDa8c8a5e318594
diff --git a/README.md b/README.md
index 59d85d1..8c585f4 100644
--- a/README.md
+++ b/README.md
@@ -2,8 +2,10 @@
<img src="imgs/title.svg" alt="BitDecoding-FP4" width="550">
</p>
-BitDecoding-FP4 is an accurate, high-performance **FP4 KV cache** for LLM
-decoding, integrated into vLLM as an out-of-tree `CUSTOM` backend.
+BitDecoding-FP4 delivers accurate, high-performance LLM decoding through
+**algorithm-system co-design**, combining **FP4 KV-cache quantization** with
+native FP4 attention kernels. It integrates into vLLM as an out-of-tree `CUSTOM`
+backend.
## Loss-less Accuracy