Fix CLS and SEP ids

server/model_api.py

@@ -52,7 +52,8 @@ class ModelDetails:
         # For GPT2, add the beginning of sentence token to the input. Note that this will work on all models but XLM
         bost = self.tok.bos_token_id
         clst = self.tok.cls_token_id
-        if (bost is not None) and (bost != clst) and add_special_tokens:
+        sept = self.tok.sep_token_id
+        if (bost is not None) and (bost != clst) and (clst != sept) and add_special_tokens:
             ids.insert(0, bost)
 
         inputs = self.tok.prepare_for_model(ids, add_special_tokens=add_special_tokens, return_tensors="pt")