Increase the max new tokens

app.py

@@ -5,7 +5,7 @@ from huggingface_hub import InferenceClient
 client = InferenceClient("mistralai/Mixtral-8x7B-Instruct-v0.1")
 
 def system_instructions(question_difficulty, tone, topic):
-    return f"""<s> [INST] Your are a great teacher and your task is to create 10 questions with 4 choices with a {question_difficulty} difficulty in a {tone} tone about {topic}, then create an answers. Index in one JSON format, the questions as "Q#":"" to "Q#":"", the four choices as "Q#:C1":"" to "Q#:C4":"", and the answers as "A#":"Q#:C#" to "A#":"Q#:C#". [/INST]"""
+    return f"""<s> [INST] Your are a great teacher and your task is to create 10 questions with 4 choices with a {question_difficulty} difficulty in a {tone} tone about {topic}, then create an answers. Index in JSON format, the questions as "Q#":"" to "Q#":"", the four choices as "Q#:C1":"" to "Q#:C4":"", and the answers as "A#":"Q#:C#" to "A#":"Q#:C#". [/INST]"""
 
 
 with gr.Blocks(title="Quizmona", theme=gr.themes.Default(primary_hue="green", secondary_hue="green"), css="style.css") as demo:
@@ -46,7 +46,7 @@ with gr.Blocks(title="Quizmona", theme=gr.themes.Default(primary_hue="green", se
 
         generate_kwargs = dict(
             temperature=0.1,
-            max_new_tokens=1024,
+            max_new_tokens=2048,
             top_p=0.95,
             repetition_penalty=1.0,
             do_sample=True,