'db_resnet50' can only be specified during training, but not during loading?

[SmilingPear]

I trained with "cnstd train -m db_resnet50 -i {} --train-config-fp {}". And obtained the weight. BUT WHEN I want to use this checkpoint with

"""
cnstd predict
-m "db_resnet50"
-b "pytorch"
-p "${MODEL_CHECKPOINT_PATH}"
......
"""

I got this：

Error: Invalid value for '-m' / '--model-name': 'db_resnet50' is not one of 'ch_PP-OCRv2_det', 'ch_PP-OCRv3_det', 'ch_PP-OCRv4_det', 'ch_PP-OCRv4_det_server', 'ch_PP-OCRv5_det', 'ch_PP-OCRv5_det_server', 'db_mobilenet_v3', 'db_mobilenet_v3_small', 'db_resnet18', 'db_resnet34', 'db_shufflenet_v2', 'db_shufflenet_v2_small', 'en_PP-OCRv3_det'.

why?