Update app.py

app.py

@@ -28,6 +28,8 @@ def predict(message, history):
     
     generation_config = GenerationConfig(
       penalty_alpha=0.6,
+      early_stopping=True, 
+      num_beams=4,
       do_sample=True,
       top_k=5,
       temperature=0.5,
@@ -39,7 +41,7 @@ def predict(message, history):
     
     outputs = model.generate(**inputs,streamer=streamer, generation_config=generation_config)
     response = tokenizer.decode(outputs[0], skip_special_tokens=True)
-    return response[len(formatted_prompt(message))-1:]
+    return response[len(formatted_prompt(message)):]
 
     # partial_message  = ""
     # for i in response: