Update app.py

app.py

@@ -267,8 +267,8 @@ def http_bot(state, temperature, top_p, max_new_tokens, speaker):
             if tts_format not in prompt and chat_format not in prompt:
                 state.messages[-1][-1] = output + "▌"
             else:
-                state.messages[-1][-1] = output + "▌" # TODO
-                # state.messages[-1][-1] = "▌"
+                # state.messages[-1][-1] = output + "▌"
+                state.messages[-1][-1] = "▌"
                 # state.messages[-1][-1] = "[😁 GENERATING AUDIO {}%...]".format(round(output.count("<|speech_") / max_new_tokens * 100, 1)) + "\n" + output + "▌"
             yield (state, state.to_gradio_chatbot_public()) + (disable_btn,) * 2
     except Exception as e:
@@ -345,7 +345,8 @@ title_markdown = ("""
     <h2 style="margin: 10px 0;">📃 <a href="https://arxiv.org/abs/2409.18042" style="font-weight: 400;">Paper</a> | 💻 <a href="https://github.com/emova-ollm/EMOVA" style="font-weight: 400;">Code</a> | 🤗 <a href="https://huggingface.co/Emova-ollm" style="font-weight: 400;">HuggingFace</a> | 🌐 <a href="https://emova-ollm.github.io/" style="font-weight: 400;">Website</a></h2>
     <p  style="margin: 20px 0;">
       <strong>1. To chat with EMOVA, upload images, enter texts or record audios and then do not forget to <mark>Click 💬 Chat Button</mark> ^v^!</strong><br/>
-      <strong>2. Heighten the <code>Max output tokens</code> if necessary to talk longer with EMOVA.</strong>
+      <strong>2. Heighten the <code>Max output tokens</code> if necessary to talk longer with EMOVA.</strong><br/>
+      <strong>3. Decrease the <code>Max output tokens</code> if <span style='color:red'>GPU task aborted</span> error is reported.</strong>
     </p>
   </div>
 </div>