Update fn.py

fn.py

@@ -206,7 +206,9 @@ def chat(message, history = [], instruction = None, args = {}):
                 # gradioは常に全文を返して欲しい
                 yield model_output
 
-    return model.generate(**generate_kwargs)
+    outputs = model.generate(**generate_kwargs)
+    content = tokenizer.decode(outputs, skip_special_tokens=True)
+    return content
 
 def apply_template(messages):
     global tokenizer, cfg