Can it work with phone number ?

[KaiserImmortal]

I trained the model with billboards but when I inference it, it doesn't work well with sequences of numbers or phone numbers. Can you help me? Thanks very much.

[simplify23]

i think you can refer to this #10. Set phone numbers as your dict.

[KaiserImmortal]

i think you can refer to this #10. Set phone numbers as your dict.
@simplify23 Sorry, you seem to be misunderstanding my question. In my dict.txt there are numeric characters from 0 to 9. I would like to ask you if your model can work with phone number sequence strings?. When I inference, it only recognizes the first numbers of the sequence, the second part is generated or incorrect. Although it still detects enough numeric characters, it fails to recognize the last numbers in the number sequence. Thanks

[KaiserImmortal]

Example:
GT: 0123526810
Pre: 0123456789

[simplify23]

Our model can fully recognize numbers, even with the previous dictionary. You can fine-tune the pre-trained model for debugging to determine where the error occurred.

[KaiserImmortal]

Our model can fully recognize numbers, even with the previous dictionary. You can fine-tune the pre-trained model for debugging to determine where the error occurred.

With 250 characters both numbers and letters, which part of the model should I fine tune. Can you suggest?

[simplify23]

Example: GT: 0123526810 Pre: 0123456789

I guess you need more training datasets. If you just need numbers, you can predict step by step with pre-train model like <s> -> <s>, 0 -> .... -> <s> 0,,1,2,3,4,5,5,6,7,8,9,<end>. I hope it can help you.

[KaiserImmortal]

Example: GT: 0123526810 Pre: 0123456789

I guess you need more training datasets. If you just need numbers, you can predict step by step with pre-train model like <s> -> <s>, 0 -> .... -> <s> 0,,1,2,3,4,5,5,6,7,8,9,<end>. I hope it can help you.

Thank you very much, i just inference with pretrained model and it work well with phone numbers. I will make more dataset and retrain