[pretrain] read text task data format quesiton

[yysirs]

pretrain read text task use english data

{\"text_sequence\": \"HANDEL SDN BHD\"}

if i use chinese data, the data format reasonable as follows:

{\"text_sequence\": \"合 同 号 : 1 2 3 \"}

or

{\"text_sequence\": \"合同号:123\"}

is the first suitable or the second? @logan-markewich @gwkrsrch

[logan-markewich]

I would say use whichever one more closely matches what is written on the document image.

If the text on the image is spaced out like the first Chinese text, then use that one. If not, use the second one.