Update app.py

app.py

@@ -318,7 +318,9 @@ def http_chat_bot(state, temperature, top_k, top_p, max_new_tokens):
     return (state, state.to_gradio_chatbot()) + (enable_btn,) * 2
 
 
-@spaces.GPU
+
+@torch.inference_mode()
+@spaces.GPU(duration=120)  # Specify a duration to avoid timeout
 def http_gen_edit_bot(state, temperature, top_k, top_p, image_gen_temperature,
                       image_gen_top_k, image_gen_top_p, max_output_tokens,
                       llm_cfg_scale, resolution_wh, use_diffusion, diffusion_cfg_scale, diffusion_num_inference_steps):
@@ -559,18 +561,17 @@ title_markdown = """
   <div>
     <h1 style="margin: 0;"> ILLUME+: Illuminating Unified MLLM with Dual Visual Tokenization and Diffusion Refinement</h1>
     <h2 style="margin: 10px 0;">
-      Links:
       <a href="https://arxiv.org/abs/2504.01934" target="_blank" rel="noopener noreferrer">Paper</a> |
       <a href="https://github.com/illume-unified-mllm/ILLUME_plus" target="_blank" rel="noopener noreferrer">Code</a> |
-      <a href="#" target="_blank" rel="noopener noreferrer">Model</a> |
+      <a href="https://huggingface.co/ILLUME-MLLM/illume_plus-qwen2_5-3b-hf" target="_blank" rel="noopener noreferrer">Model</a> |
       <a href="https://illume-unified-mllm.github.io/" target="_blank" rel="noopener noreferrer">Project Page</a>
     </h2>
     <ul style="margin: 20px 0; padding-left: 20px;">
       <li><strong>1.</strong> Enter text and/or upload an image.</li>
       <li><strong>2.</strong> Click the 💬 <strong>Chat</strong> button for image inputted conversations</li>
       <li><strong>3.</strong> Click the 🖼️ <strong>Generate</strong> for image generation and image editing.</li>
-      <li><strong>5.</strong> (Optional) Enable Diffusion Decoder for image super resolution decoding. 
-      <li><strong>4.</strong> Adjust generation parameters if needed. 
+      <li><strong>4.</strong> (Optional) Enable Diffusion Decoder for image super resolution decoding. 
+      <li><strong>5.</strong> Adjust generation parameters if needed. 
         <br/><strong>💡 Tip 1:</strong> For better image generation quality, we recommend setting <code>temperature = 1.0</code>, <code>top_k = 2048</code>, <code>top_p = 1.0</code>, <code>llm_cfg = 2.0</code>.    
         <br/><strong>💡 Tip 2:</strong> For better image editing quality, we recommend setting <code>temperature = 0.7</code>, <code>top_k = 512</code>, <code>top_p = 0.8</code>, <code>llm_cfg = 1.5</code>.
         <br/><strong>💡 Tip 3:</strong> For diffusion decoder, CFG scale of 1.5 or 2.0 is enough.
@@ -960,5 +961,4 @@ if __name__ == "__main__":
         api_open=False
     ).launch(
         share=args.share,
-        server_name="0.0.0.0"  # Allow network access if not using --share
     )