Update app.py

app.py

@@ -93,11 +93,11 @@ def generate(prompt, temperature=0.9, max_new_tokens=256, top_p=0.95, repetition
         output = prompt
 
     for response in stream:
-        output += response.token.text
+        if fim_mode and response.token.text =="<|endoftext|>":
+            output += (suffix + response.token.text)
+        else:
+            output += response.token.text
         yield output
-    
-    if fim_mode:
-        output += suffix
     return output