Update README.md

README.md

@@ -43,7 +43,7 @@ to get the pruned model.
 # Benchmark Results
 
 **Benchmark Evaluation**:
-The model follows  evaluation and perform zero-shot task classification on 5 common sense
+The model follows the original paper's evaluation and perform zero-shot task classification on 5 common sense
 reasoning datasets that doesn't require foreign code to load:
 
 | Model                        | BoolQ  | HellaSwag | ARC-e  | ARC-c  | OBQA  | Average Accuracy  |