change default token length

app.py

@@ -64,10 +64,9 @@ def predict_example(temperature, top_p, max_new_token, keep_last_n_paragraphs, i
 if __name__ == '__main__':
     parser = argparse.ArgumentParser()
     parser.add_argument('--temperature', type=float, default=0.0, help='temperature for the llm model')
-    parser.add_argument('--max_new_tokens', type=int, default=128, help='max number of new tokens to generate')
+    parser.add_argument('--max_new_tokens', type=int, default=256, help='max number of new tokens to generate')
     parser.add_argument('--top_p', type=float, default=1.0, help='top_p for the llm model')
     parser.add_argument('--top_k', type=int, default=40, help='top_k for the llm model')
-    parser.add_argument('--num_beams', type=int, default=4, help='num_beams for the llm model')
     parser.add_argument('--keep_last_n_paragraphs', type=int, default=0, help='keep last n paragraphs in the memory')
     args = parser.parse_args()