Update README.md

README.md

@@ -12,4 +12,4 @@ widget:
 A further fine-tuned version of Locutusque/gpt2-large-conversational on MedText and pubmed_qa
 
 # Evaluation
-This model was evaluated using GPT-3.5, and it was asked medical questions. It achieved an average accuracy of 75.8%.
+This model was evaluated using GPT-3.5, and it was asked medical questions. It achieved an average accuracy of 80%.