Update app.py

app.py

@@ -17,8 +17,8 @@ from transformers import (
 from qwen_vl_utils import process_vision_info
 
 # Constants for text generation
-MAX_MAX_NEW_TOKENS = 4096
-DEFAULT_MAX_NEW_TOKENS = 3584
+MAX_MAX_NEW_TOKENS = 16384
+DEFAULT_MAX_NEW_TOKENS = 8192
 MAX_INPUT_TOKEN_LENGTH = int(os.getenv("MAX_INPUT_TOKEN_LENGTH", "4096"))
 
 device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")