Update README.md

README.md

@@ -18,9 +18,9 @@ Finally, the dataset we used to train the model contains 41,060 positive and neg
 
 # Results
 **MHC-I-EpiPred** achieved the following results:
-Training Loss (mse): 0.1044  
+Training Loss (cross-entropy loss, CEL): 0.1044  
 Training Accuracy: 98.99%   
-Evaluation Loss (mse): 0.1576   
+Evaluation Loss (cross-entropy loss, CEL): 0.1576   
 Evaluation Accuracy: 97.04%  
 Epochs: 492