Update README.md

README.md

@@ -39,7 +39,7 @@ The training process included:
 - Framework: Fine-tuning was conducted using Unsloth, enabling faster training and better memory efficiency.
 
 ## Evaluation
-The model was evaluated using the **MathQA test dataset** with **accuracy** as the primary metric. The following table compares its performance to other models:
+The model was evaluated using the **MathQA test dataset(2985 examples)** with **accuracy** as the primary metric. The following table compares its performance to other models:
 
 | Model                | Accuracy (%)  |
 |----------------------|---------------|