Update app.py

app.py

@@ -10,14 +10,19 @@ client = Groq(api_key=os.environ.get("Groq_Api_Key"))
 
 # llms
 
+MAX_SEED = np.iinfo(np.int32).max
+
+def update_max_tokens(model):
+    if model in ["llama3-70b-8192", "llama3-8b-8192", "gemma-7b-it", "gemma2-9b-it"]:
+        return gr.update(maximum=8192)
+    elif model == "mixtral-8x7b-32768":
+        return gr.update(maximum=32768)
+
 def create_history_messages(history):
     history_messages = [{"role": "user", "content": m[0]} for m in history]
     history_messages.extend([{"role": "assistant", "content": m[1]} for m in history])
     return history_messages
 
-
-MAX_SEED = np.iinfo(np.int32).max
-
 def generate_response(prompt, history, model, temperature, max_tokens, top_p, seed):
     messages = create_history_messages(history)
     messages.append({"role": "user", "content": prompt})
@@ -257,7 +262,19 @@ with gr.Blocks() as demo:
     with gr.Tabs():
         with gr.TabItem("LLMs"):
             with gr.Row():
-                with gr.Column():
+                with gr.Column(scale=1, min_width=400): 
+                    chatbot = gr.ChatInterface(
+                        fn=generate_response,
+                        chatbot=None,
+                        additional_inputs=[
+                            model,
+                            temperature,
+                            max_tokens,
+                            top_p,
+                            seed,
+                        ],
+                    )
+                with gr.Column(scale=1, min_width=250):
                     model = gr.Dropdown(
                         choices=[
                             "llama3-70b-8192",
@@ -279,7 +296,7 @@ with gr.Blocks() as demo:
                     )
                     max_tokens = gr.Slider(
                         minimum=1,
-                        maximum=32192,
+                        maximum=8192,
                         step=1,
                         value=4096,
                         label="Max Tokens",
@@ -296,12 +313,7 @@ with gr.Blocks() as demo:
                     seed = gr.Number(
                         precision=0, value=42, label="Seed", info="A starting point to initiate generation, use 0 for random"
                     )
-                with gr.Column():
-                    chatbot_ui = gr.ChatInterface(
-                        fn=generate_response,
-                        chatbot=None,
-                        additional_inputs=[model, temperature, max_tokens, top_p, seed],
-                    )
+                    model.change(update_max_tokens, inputs=[model], outputs=max_tokens)
         with gr.TabItem("Speech To Text"):
             with gr.Tabs():
                 with gr.TabItem("Transcription"):