Update README.md

README.md

@@ -46,6 +46,9 @@ The model is fine-tuned for 4000 updates/steps on multilingual librispeech Spani
 - Zero-shot                      - 6.3 (CV11 test)
 - Fine-tune MLS spanish train    - 8.38 (CV11 test)
 
+When the model is fine-tuned on specific dataset, the model loose its ability to generalise across datasets.
+Here the model is fine-tuned on MLS Spanish and evaluated on CV11 Spanish test. We can observe the drop in performance on CV11 test data.
+
 ## Intended uses & limitations
 
 More information needed