init

app.py

@@ -47,7 +47,7 @@ def start_worker(model_path: str, bits=4):
         "--model-path",
         model_path,
         "--model-name",
-        'llava-v1.5-UGround-4bit',
+        'llava-UGround-v1-4bit',
         "--use-flash-attn",
     ]
     if bits != 16:

llava/conversation.py

@@ -196,7 +196,7 @@ class Conversation:
                     img_str = f'<img src="data:image/jpeg;base64,{img_b64_str}" alt="user upload image" />'
                     msg=img_str
                     if not extra_coordinates:
-                        msg=f"The element is at {extra_coordinates} on the screen: "  +msg
+                        ret.append(["The element is at {extra_coordinates} on the screen:",None])
                     ret.append([msg, None])
                 break
         return ret

llava/serve/gradio_web_server.py

@@ -447,7 +447,7 @@ def build_demo(embed_mode, cur_dir=None, concurrency_count=1):
                         interactive=True,
                         show_label=False,
                         container=False)
-                # model_selector="llava-v1.5-UGround-4bit"
+                # model_selector="llava-UGround-v1-4bit"
 
                 imagebox = gr.Image(type="pil")
                 image_process_mode = gr.Radio(

pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "uground_demo_test"
-version = "3.9"
+version = "4.0"
 description = "Navigating the Digital World as Humans Do: Universal Visual Grounding for GUI Agents"
 readme = "README.md"
 requires-python = ">=3.8"