Update README.md

README.md

@@ -60,8 +60,14 @@ The base model used is the Vision Transformer `vit-base-patch16-224-in21k`, whic
 The model was trained on a curated dataset of UAE company logos as well as others of international companies. The dataset consists of thousands of images across various brands to ensure robustness and accuracy.
 
 ### Performance
-The model achieved high accuracy on a held-out validation set, indicating strong performance in classifying UAE company logos. Detailed performance metrics (accuracy, precision, recall, F1-score) can be provided upon request.
-
+***** eval metrics *****
+-  epoch                   =       20.0
+-  eval_accuracy           =     0.9761
+-  eval_loss               =     0.1193
+-  eval_runtime            = 0:00:20.51
+-  eval_samples_per_second =    268.951
+-  eval_steps_per_second   =      8.432
+  
 ## How to Use
 To use the model for inference, you can load it using the `transformers` library from Hugging Face:
 
@@ -154,7 +160,7 @@ print(model.config.id2label[predicted_label])
 
 
 ### Limitations and Biases
-- The model is specifically trained on UAE company logos and may not perform well on logos from companies outside the UAE.
+- The model is specifically trained on UAE company logos and a few from Outside the UAE, it may not perform well on logos from numerous other large or small companies.
 - The model's performance is contingent upon the quality and diversity of the training dataset.
 - Potential biases in the training data can lead to biases in model predictions.