Update README.md

README.md

@@ -9,7 +9,7 @@ You could check our project page at [🏠PromptGuard HomePage](https://prompt-gu
 
 In the future, we will release our training datasets. 
 
-# Usage
+# Inference
 A simple use case of our model is:
 ```python
 from diffusers import StableDiffusionPipeline
@@ -35,3 +35,5 @@ prompt_with_system = origin_prompt + " " + token1 + " " + token2 + ...
 image = pipe(prompt).images[0]
 image.save("example.png")
 ```
+
+To get a better balance between unsafe content moderation and benign content preservation, we recommend you to load Sexual, Political and Disturbing these three safe embeddings.