converted to tensor

model.py

@@ -63,7 +63,8 @@ class VirTexModel():
             subreddit_tokens = torch.tensor(subreddit_tokens, device=self.device).long()
             
         if prompt is not "":
-            cap_tokens = self.tokenizer.encode(prompt) 
+            cap_tokens = self.tokenizer.encode(prompt)
+            cap_tokens = torch.tensor(cap_tokens, device=self.device).long()
             subreddit_tokens = torch.cat([subreddit_tokens, cap_tokens])
             
         predictions: List[Dict[str, Any]] = []