diff --git a/lm_eval/models/gptj.py b/lm_eval/models/gptj.py
index 398ae03053..92ea18bc34 100644
--- a/lm_eval/models/gptj.py
+++ b/lm_eval/models/gptj.py
@@ -39,6 +39,10 @@ def __init__(
         # if gpus > 1:
         #     self.gptj = nn.DataParallel(self.gptj)
 
+    @property
+    def eot_token(self):
+        return self.tokenizer.eos_token
+
     @property
     def eot_token_id(self):
         # we use EOT because end of *text* is more accurate for what we're doing than end of *sentence*