Update app.py

app.py

@@ -19,7 +19,7 @@ generator = pipeline(
 # Create text streamer
 streamer = TextStreamer(generator.tokenizer, skip_prompt=True)
 
-def generate_response(prompt: str, max_new_tokens: int = 64000):
+def generate_response(prompt: str, max_new_tokens: int = 4096):
     try:
         messages = [{"role": "user", "content": prompt}]
         output = generator(messages, max_new_tokens=max_new_tokens, do_sample=False, streamer=streamer)