chore: switch to Qwen2.5-VL-3B-Instruct for OCR

- Upgrade from Qwen2-VL-2B to Qwen2.5-VL-3B
- Improved model with better performance

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude <noreply@anthropic.com>

app.py

@@ -10,8 +10,8 @@ from PIL import Image
 from transformers import Qwen2VLForConditionalGeneration, AutoProcessor
 from qwen_vl_utils import process_vision_info
 
-# Qwen2-VL 모델 ID
-MODEL_ID = "Qwen/Qwen2-VL-2B-Instruct"
+# Qwen2.5-VL 모델 ID
+MODEL_ID = "Qwen/Qwen2.5-VL-3B-Instruct"
 
 
 def _extract_assistant_content(decoded: str) -> str:
@@ -298,7 +298,7 @@ with gr.Blocks(theme=gr.themes.Soft(), css=CUSTOM_CSS) as demo:
     ---
 
     **ℹ️ OCR 모델**
-    - Qwen2-VL-2B-Instruct - 최첨단 비전-언어 모델 기반 OCR (GPT-4o 수준)
+    - Qwen2.5-VL-3B-Instruct - 최첨단 비전-언어 모델 기반 OCR (GPT-4o 수준)
     - 한국어, 영어 등 다국어 지원
     """)