Update README.md

README.md

@@ -45,7 +45,7 @@ Baichuan-M2 incorporates three core technical innovations: First, through the **
 
 ### General Performance
 
-| Benchmark | Baichuan-M2-32B | Qwen3-32B |
+| Benchmark | Baichuan-M2-32B | Qwen3-32B (Thinking) |
 |-----------|-----------------|-----------|
 | AIME24 | 83.4 | 81.4 |
 | AIME25 | 72.9 | 72.9 |
@@ -75,10 +75,19 @@ For deployment, you can use `sglang>=0.4.6.post1` or `vllm>=0.9.0` or to create
     ```shell
     python -m sglang.launch_server --model-path baichuan-inc/Baichuan-M2-32B-GPTQ-Int4 --reasoning-parser qwen3
     ```
+To turn on kv cache FP8 quantization:
+    ```shell
+    python -m sglang.launch_server --model-path baichuan-inc/Baichuan-M2-32B-GPTQ-Int4 --reasoning-parser qwen3 --kv-cache-dtype fp8_e4m3 --attention-backend flashinfer
+    ```
+
 - vLLM:
     ```shell
     vllm serve baichuan-inc/Baichuan-M2-32B-GPTQ-Int4  --reasoning-parser qwen3
     ```
+To turn on kv cache FP8 quantization:
+    ```shell
+    vllm serve baichuan-inc/Baichuan-M2-32B-GPTQ-Int4  --reasoning-parser qwen3 --kv_cache_dtype fp8_e4m3
+    ```
 
 ## MTP inference with SGLang