Update README.md

README.md

@@ -27,6 +27,8 @@ Today, I am excited to announce the release of two models specifically trained t
 
 <h2>Model Evaluation</h2>
 
+We evaluated this model on the [VLMU benchmark](https://vmlu.ai/) and achieved an accuracy of 45.13. However, this benchmark is not the focus of our current efforts. We believe it will be very difficult for language models with fewer than 13 billion parameters to retain enough knowledge to answer questions across diverse user contexts, especially for smaller models with under 3 billion parameters. For the model to effectively handle real-world business scenarios and avoid hallucinations, it is almost essential to supplement knowledge from external sources (through RAG). Therefore, we developed this model with a primary focus on optimizing its RAG capabilities. Internal testing is currently underway and will be updated soon.
+
 Will be updated in the coming days.