Update README.md

README.md

@@ -43,4 +43,4 @@ Every tranch was downloaded, this is not the ~1B example ML subset from https://
 
 The dataset was entirely shuffled then split into 80%/10%/10% splits for train/val/test.
 
-A file vocab.json is in the root of the reposity that contains all of the SELFIES tokens found in the data, with [START], [STOP], and [PAD] added.
+A file vocab.csv is in the root of the reposity that contains all of the SELFIES tokens found in the data, with [START], [STOP], and [PAD] added.