Update app.py

app.py

@@ -29,12 +29,12 @@ def predict(message, history):
     generation_config = GenerationConfig(
       penalty_alpha=0.6,
       early_stopping=True, 
-      num_beams=2,
+      num_beams=4,
       do_sample=True,
       top_k=5,
       temperature=0.5,
       repetition_penalty=1.2,
-      # max_new_tokens=64,
+      max_new_tokens=128,
       eos_token_id=tokenizer.eos_token_id,
       pad_token_id=tokenizer.eos_token_id
   )