diff --git a/README.md b/README.md
index 47b5555..ab906cf 100644
--- a/README.md
+++ b/README.md
@@ -33,6 +33,8 @@ the training here:
 
 We also provide cased and uncased models that aŕe using a larger vocab size (128k instead of 32k).
 
+A detailed cheetsheat of how the models were trained, can be found [here](CHEETSHEAT.md).
+
 ## *DistilBERTurk*
 
 The distilled version of a cased model, so called *DistilBERTurk*, was trained