Update app.py

app.py

@@ -16,7 +16,7 @@ def format_math(text):
     text = text.replace(r"\(", "$").replace(r"\)", "$")
     return text
 
-@spaces.GPU(duration=60)
+#@spaces.GPU(duration=60)
 def generate_response(user_message, max_tokens, temperature, top_p, history_state):
     if not user_message.strip():
         return history_state, history_state
@@ -45,9 +45,9 @@ def generate_response(user_message, max_tokens, temperature, top_p, history_stat
         "max_new_tokens": int(max_tokens),
         "do_sample": True,
         "temperature": temperature,
-        "top_k": 50,  # Fixed default
+        "top_k": 50,
         "top_p": top_p,
-        "repetition_penalty": 1.0,  # Fixed default
+        "repetition_penalty": 1.0,
         "pad_token_id": tokenizer.eos_token_id,
         "streamer": streamer,
     }
@@ -102,37 +102,22 @@ with gr.Blocks(theme=gr.themes.Soft()) as demo:
     with gr.Row():
         with gr.Column(scale=1):
             gr.Markdown("### Settings")
-            max_tokens_slider = gr.Slider(
-                minimum=6144,
-                maximum=32768,
-                step=1024,
-                value=16384,
-                label="Max Tokens"
-            )
+            max_tokens_slider = gr.Slider(minimum=6144, maximum=32768, step=1024, value=16384, label="Max Tokens")
             with gr.Accordion("Advanced Settings", open=False):
-                temperature_slider = gr.Slider(
-                    minimum=0.1,
-                    maximum=2.0,
-                    value=0.6,
-                    label="Temperature"
-                )
-                top_p_slider = gr.Slider(
-                    minimum=0.1,
-                    maximum=1.0,
-                    value=0.95,
-                    label="Top-p"
-                )
+                temperature_slider = gr.Slider(minimum=0.1, maximum=2.0, value=0.6, label="Temperature")
+                top_p_slider = gr.Slider(minimum=0.1, maximum=1.0, value=0.95, label="Top-p")
 
         with gr.Column(scale=4):
             chatbot = gr.Chatbot(label="Chat", type="messages")
+            user_input = gr.Textbox(
+                label="User Input",
+                placeholder="Type your question here...",
+                lines=2,
+                scale=1,
+                full_width=True
+            )
             with gr.Row():
-                user_input = gr.Textbox(
-                    label="User Input",
-                    placeholder="Type your question here...",
-                    scale=3
-                )
                 submit_button = gr.Button("Send", variant="primary", scale=1)
-                #stop_button = gr.Button("Stop", variant="stop", scale=1, interactive=True)
                 clear_button = gr.Button("Clear", scale=1)
             gr.Markdown("**Try these examples:**")
             with gr.Row():
@@ -172,4 +157,4 @@ with gr.Blocks(theme=gr.themes.Soft()) as demo:
         outputs=user_input
     )
 
-demo.launch(share=True, ssr_mode=False)
+demo.launch(share=True, ssr_mode=False)