Update app.py

app.py

@@ -111,16 +111,16 @@ def generate_response(instruction, model_path, progress=gr.Progress()):
             do_sample=True
         )
         
-    full_response = response_tokenizer.decode(output[0], skip_special_tokens=True)
+        full_response = response_tokenizer.decode(output[0], skip_special_tokens=True)
     
-    clean_response = full_response.replace(f"<|user|>\n{instruction}\n<|assistant|>", "").strip()
+        clean_response = full_response.replace(f"<|user|>\n{instruction}\n<|assistant|>", "").strip()
     
-    if "<|user|>" in clean_response:
-        clean_response = clean_response.split("<|user|>")[0].strip()
+        if "<|user|>" in clean_response:
+            clean_response = clean_response.split("<|user|>")[0].strip()
     
-    for token in ["<user>", "User:", "Human:"]:
-        if token in clean_response:
-            clean_response = clean_response.split(token)[0].strip()
+        for token in ["<user>", "User:", "Human:"]:
+            if token in clean_response:
+                clean_response = clean_response.split(token)[0].strip()