diff --git a/index.md b/index.md
index b9c35b9..223e296 100644
--- a/index.md
+++ b/index.md
@@ -4,16 +4,9 @@
# Robust Prompt Optimization for Defending Language Models Against Jailbreaking Attacks
-
-
-
-
Fig.1 GPT-4 safety filters can be bypassed by jailbreaks!
-
-
-
-
Fig.2 RPO enforces harmless responses even after jailbreaks
-
-
+| ![GPT-4](figures/gpt4.png) | ![RPO](figures/gpt4_rpo.png) |
+|:--:|:--:|
+| *Fig.1 GPT-4 safety filters can be bypassed by jailbreaks!* | *Fig.2 RPO enforces harmless responses even after jailbreaks* |
## Abstract