Update README.md

README.md

@@ -1,6 +1,6 @@
 This model was pretrained on the bookcorpus dataset using knowledge distillation.
 
-The particularity of this model is that even though it shares the same architecture as BERT, it has a hidden size of 384 (half the hidden size of BERT) and 6 attention heads (hence the same head size of BERT).
+The particularity of this model is that even though it shares the same architecture as BERT, it has a hidden size of 256 (a third of the hidden size of BERT) and 4 attention heads (hence the same head size of BERT).
 
 The weights of the model were initialized by pruning the weights of bert-base-uncased.