Using translate.py to predict file is ok,but using translate.py to predict single text with params --text has a problem. #2
Description
ValueError: Dimensions must be equal, but are 36 and 256 for 'model/Transformer/encode/add_pos_encoding/add' (op: 'Add') with input shapes: [36,256], [256,256].