Update README.md

README.md

@@ -148,7 +148,7 @@ The adaptation procedure is explained in [this blog post](https://medium.com/@mp
 
 The training corpus consists of 26B tokens of several corpora gathered from web crawlings and public domain data.
 
-| Dataset             | Language | Tokens (per-epoch) | Epochs       |
+| Dataset             | Language | Words (per-epoch) | Epochs       |
 |---------------------|----------|--------------------|--------------|
 | Wikipedia           | en       |           2169.97M |  1.428144485 |
 | C4_es               | es       |          53709.80M | 0.1049686196 |