1

app.py

@@ -101,6 +101,7 @@ def predict(history, max_length, top_p, temperature, img_path, platform_str, for
         "do_sample": True if temperature > 0.0 else False,
         "top_p": top_p,
         "temperature": temperature,
+        "top_k": 1,
     }
     t = Thread(target=model.generate, kwargs=generate_kwargs)
     t.start()
@@ -186,11 +187,12 @@ def main():
     with gr.Blocks(analytics_enabled=False) as demo:
         gr.HTML("<h1 align='center'>CogAgent-9B-20241220 Demo</h1>")
         gr.HTML(
-            "<p align='center' style='color:red;'>This Demo is for learning and communication purposes only. Users must assume responsibility for the risks associated with AI-generated planning and operations.</p>")
-
+            "<p align='center' style='color:red;'>This demo is for learning and communication purposes only. Users must assume responsibility for the risks associated with AI-generated planning and operations.</p>"
+            "<p align='left' style='color:black;'>1. Upload an image. 2. Provide your instructions to CogAgent. 3. Wait for CogAgent to return specific operations, and if there are bounding boxes (Bbox), they will be displayed in the image area on the right.</p>"
+        )
         with gr.Row():
             img_path = gr.Image(label="Upload a Screenshot", type="filepath", height=400)
-            output_img = gr.Image(type="filepath", label="Annotated Image", height=400, interactive=False)
+            output_img = gr.Image(type="filepath", label="Annotated Image(If Bbox Return)", height=400, interactive=False)
 
         with gr.Row():
             with gr.Column(scale=2):