Update README.md

README.md

@@ -19,7 +19,7 @@ This repository contains pretrained ProtoViT models for interpretable image clas
 
 ## Model Description
 
-ProtoViT combines Vision Transformers with prototype-based learning to create models that are both highly accurate and interpretable. Rather than functioning as a black box, ProtoViT learns interpretable prototypes that help explain its classification decisions through visual similarities.
+[ProtoViT](https://github.com/Henrymachiyu/ProtoViT) combines Vision Transformers with prototype-based learning to create models that are both highly accurate and interpretable. Rather than functioning as a black box, ProtoViT learns interpretable prototypes that explain its classification decisions through visual similarities.
 
 ### Supported Architectures
 
@@ -55,6 +55,11 @@ All models were trained and evaluated on the CUB-200-2011 fine-grained bird spec
 - numpy
 - pillow
 
+## Limitations and Bias
+- Data Bias: These models are trained on CUB-200-2011, which may not generalize well to images outside this dataset.
+- Resolution Constraints: The models are trained at a resolution of 224×224; higher or lower resolutions may impact performance.
+- Location Misalignment: Same as the other CNN based models, these models are not perfectly immune to location misalignment under adversarial attack. 
+
 ## Citation
 
 If you use this model in your research, please cite: