Update README.md

README.md

@@ -276,7 +276,7 @@ prediction = rr.rerank_batch(candidates['test'], ce_dataset['test'])
 
 #### Note on Memory Usage
 In most examples and benchmarks, we use 64 candidates as a batch size for the cross-encoder, which usually fit into 48GB of GPU memory. 
-If you encounter memory issue, you can try reducing this number and/or using a smaller BERT model. See: [Issue #22](https://github.com/hpi-dhc/xmen/issues/22#issuecomment-1827648900)
+If you encounter memory issues, you can try reducing this number and/or using a smaller BERT model. See: [Issue #22](https://github.com/hpi-dhc/xmen/issues/22#issuecomment-1827648900)
 
 ### Pre-trained Cross-encoders