Update app.py

app.py

@@ -42,6 +42,10 @@ def load_models(inp):
         print(models[inp])
     client_z.clear()
     client_z.append(InferenceClient(models[inp]))
+
+    if "mistralai" in models[inp]:
+        
+    
     return gr.update(label=models[inp])
 
 def format_prompt(message, history):
@@ -61,7 +65,7 @@ agents =[
 
 temperature=0.9
 max_new_tokens=256
-max_new_tokens2=10480
+max_new_tokens2=4000
 top_p=0.95
 repetition_penalty=1.0,
 
@@ -77,7 +81,7 @@ def compress_history(formatted_prompt):
 
     generate_kwargs = dict(
         temperature=temperature,
-        max_new_tokens=10480,
+        max_new_tokens=1048,
         top_p=0.95,
         repetition_penalty=1.0,
         do_sample=True,
@@ -352,7 +356,9 @@ with gr.Blocks() as app:
         submit_b = gr.Button()
         stop_b = gr.Button("Stop")
         clear = gr.ClearButton([msg, chatbot])
-    m_choice=gr.Dropdown(label="Models",type='index',choices=[c for c in models],value=models[0],interactive=True)
+    with gr.Row():
+        m_choice=gr.Dropdown(label="Models",type='index',choices=[c for c in models],value=models[0],interactive=True)
+        tokens = gr.Slider(label="Max new tokens",value=1600,minimum=0,maximum=8000,step=64,interactive=True, visible=True,info="The maximum number of tokens")
 
     sumbox=gr.Textbox("Summary", max_lines=100)
     with gr.Column():
@@ -363,8 +369,8 @@ with gr.Blocks() as app:
     m_choice.change(load_models,m_choice,[chatbot])
     app.load(load_models,m_choice,[chatbot]).then(load_html,None,html)
     
-    sub_b = submit_b.click(generate, [msg,chatbot],[msg,chatbot,sumbox,sum_out_box,hist_out_box,html])
-    sub_e = msg.submit(generate, [msg, chatbot], [msg, chatbot,sumbox,sum_out_box,hist_out_box,html])
+    sub_b = submit_b.click(generate, [msg,chatbot,tokens],[msg,chatbot,sumbox,sum_out_box,hist_out_box,html])
+    sub_e = msg.submit(generate, [msg, chatbot,tokens], [msg, chatbot,sumbox,sum_out_box,hist_out_box,html])
     stop_b.click(None,None,None, cancels=[sub_b,sub_e])
     
 app.queue(default_concurrency_limit=20).launch()