Add legends

src/display/about.py

@@ -62,8 +62,6 @@ Examples of these vulnerabilities are shown below.
 ![DecodingTrust Overview](https://decodingtrust.github.io/overview.png)
 
 
-We normalize the score of each perspective as 0-100, and these scores are the higher the better.
-
 ## Reproducibility
 Please follow the following instructions to install DecodingTrust
 
@@ -133,6 +131,15 @@ singularity exec --nv --bind /path/on/host:/path/in/container decoding-trust-v1.
 
 To run our evaluations, checkout our [tutorial](https://github.com/AI-secure/DecodingTrust/blob/main/Tutorial.md).
 
+## Icons
++ 🟢 Pretrained Model: Base models trained on a given corpora
++ 🔶 Fine-tuned Model: Pretrained model fine-tuned on more data
++ ⭕ Instruction-tuned Model: Models specifically fine-tuned on datasets with task instructions
++ 🟦 RL-tuned: Models specifically fine-tuned with RLHF or DPO
+
+If there is no icon, we have not uploaded the information on the model yet, feel free to open an issue with the model
+information
+
 ## Changelog
 01/17/24: Add the initial version of the leaderboard