Update app.py

app.py

@@ -2,7 +2,7 @@ from huggingface_hub import InferenceClient
 import gradio as gr
 import random
 
-client = InferenceClient("THUDM/codegeex2-6b")
+client = InferenceClient("mistralai/Mixtral-8x7B-Instruct-v0.1")
 
 def format_prompt(message, history):
   prompt = "<s>\n"
@@ -22,7 +22,7 @@ def generate(
 
     generate_kwargs = dict(
         temperature=temperature,
-        # max_new_tokens=max_new_tokens,
+        max_new_tokens=max_new_tokens,
         top_p=top_p,
         repetition_penalty=repetition_penalty,
         do_sample=True,