Update README.md

README.md

@@ -140,9 +140,16 @@ For further testing our decoder, in addition to the testing data described above
 | **LLaMA-2 (English)**              | **0.2458**               | 0.2903          | 0.0913    | 0.1034        |
 | **LLaMA-2 Chat (English)**         | 0.2231               | **0.2959**          | 0.5546    | 0.1750        |
 
-<br>
 
 
+In comparison with other decoder of the same dimension, namely Sabiá 1.5B, Gervásio shows a superior
+or competitive performance for the tasks in PTBR, while being the sole encoder of 1.5B dimmension for the PTPT 
+variant of Portuguese and thus the state of art
+in this respect at the time of its publishing. For further evaluation data, 
+see the respective [publication](https://arxiv.org/abs/2402.18766).
+
+<br>
+
 # How to use
 
 You can use this model directly with a pipeline for causal language modeling: