Update app.py

app.py

@@ -88,7 +88,7 @@ def downsample_video(video_path):
     vidcap.release()
     return frames
 
-MODEL_ID = "kaamd/chtvctr" # "kaamd/chtvctr"  # Alternatively: "Qwen/Qwen2.5-VL-3B-Instruct" 
+MODEL_ID = "yaya-sy/chvtr" # "kaamd/chtvctr"  # Alternatively: "Qwen/Qwen2.5-VL-3B-Instruct" 
 processor = AutoProcessor.from_pretrained(MODEL_ID, trust_remote_code=True, min_pixels=256*28*28, max_pixels=1280*28*28)
 model = Qwen2_5_VLForConditionalGeneration.from_pretrained(
     MODEL_ID,
@@ -191,11 +191,11 @@ def model_inference(input_dict, history):
 
 # Option 1: Use regular Interface with streaming (recommended)
 with gr.Blocks() as demo:
-    gr.Markdown("# oolel-vision-experimental `@video-infer for video understanding`")
+    gr.Markdown("# Oolel`")
     
     chatbot = gr.Chatbot()
     msg = gr.MultimodalTextbox(
-        label="Query Input", 
+        label="Your Request", 
         file_types=["image", "video"], 
         file_count="multiple"
     )