Update README.md

README.md

@@ -88,9 +88,11 @@ with torch.no_grad():
 # 4</s>
 ```
 
-当单卡显存不足时，可以将模型切分到多个小显存GPU上
+当单卡显存不足时，可以将模型切分到多个小显存GPU上。以下是个当你有两张24GB的GPU，16GBCPU内存的例子。
+你可以将`infer_auto_device_map`的参数改成你的配置。注意这里将GPU显存少写了一点，这是为推理时中间状态预留出一部分显存。
 
-dispatch the model into multiple GPUs with smaller VRAM. 
+dispatch the model into multiple GPUs with smaller VRAM. This is an example for you have two 24GB GPU and 16GB CPU memory.
+you can change the arguments of `infer_auto_device_map` with your own setting.
 
 ```python
 import torch