How Train vistaOCR?

[Tailor2019]

Hello
@stephenrawls
When i'm trying to train this system with oflline handritten recognition database it returns me the error that :
-the the file "desc.json" loaded in this classe ""isi-vista/VistaOCR/blob/master/src/ocr_dataset.py""
with open(os.path.join(data_dir, 'desc.json'), 'r') as fh:
self.data_desc = json.load(fh)
1-Is the file "desc.json"" the ground truth of the database used and can you make me please the description of this File and how can implement it?

2-It is this instruction used to train correct and what is the value we must put to "--snapshot-prefix" 👍
python train_cnn_lstm.py --datadir /root/Downloads/data --num-lstm-layers 3 --num-lstm-units 256 --lstm-input-dim=256 --snapshot-prefix /root/Downloads/VistaOCR-master/src/models

thank in advance