Set better defaults for free-tier users: Qwen3-1.7B model, 1024 max tokens, search disabled

app.py

@@ -593,11 +593,11 @@ with gr.Blocks(title="LLM Inference with ZeroGPU") as demo:
     gr.Markdown("Interact with the model. Select parameters and chat below.")
     with gr.Row():
         with gr.Column(scale=3):
-            model_dd = gr.Dropdown(label="Select Model", choices=list(MODELS.keys()), value=list(MODELS.keys())[0])
-            search_chk = gr.Checkbox(label="Enable Web Search", value=True)
+            model_dd = gr.Dropdown(label="Select Model", choices=list(MODELS.keys()), value="Qwen3-1.7B")
+            search_chk = gr.Checkbox(label="Enable Web Search", value=False)
             sys_prompt = gr.Textbox(label="System Prompt", lines=3, value=update_default_prompt(search_chk.value))
             gr.Markdown("### Generation Parameters")
-            max_tok = gr.Slider(64, 16384, value=2048, step=32, label="Max Tokens")
+            max_tok = gr.Slider(64, 16384, value=1024, step=32, label="Max Tokens")
             temp = gr.Slider(0.1, 2.0, value=0.7, step=0.1, label="Temperature")
             k = gr.Slider(1, 100, value=40, step=1, label="Top-K")
             p = gr.Slider(0.1, 1.0, value=0.9, step=0.05, label="Top-P")