Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Train custom dataset ? #4

Open
ThorPham opened this issue Nov 19, 2021 · 4 comments
Open

Train custom dataset ? #4

ThorPham opened this issue Nov 19, 2021 · 4 comments

Comments

@ThorPham
Copy link

ThorPham commented Nov 19, 2021

I want to train in my language . What is config i must modify ?
if label_convertor['dict_type'] == 'DICT90': PAD = 92
What is pad = 92 mean .

@JiaquanYe
Copy link
Owner

I want to train in my language . What is config i must modify ? if label_convertor['dict_type'] == 'DICT90': PAD = 92 What is pad = 92 mean .

First, you should build your character dictionary, like DICT90, and use your custom dictionary in your config file.
PAD is a special symbol of MASTER, which is out of DICT90 but will be used in MASTER Decoder. Other like SOS/EOS, is also MASTER's special symbol.
PAD is mean this token will been use to pad the sequence to max_length.

@ThorPham
Copy link
Author

@JiaquanYe Thank you for reply . Can model recognition space ? I want to add space character in DICT90.

@JiaquanYe
Copy link
Owner

@JiaquanYe Thank you for reply . Can model recognition space ? I want to add space character in DICT90.

Sure, you can add space character in your custom dictionary, and apply it in the training config.

@bharatsubedi
Copy link

@ThorPham did you train this model without failure? I am facing errors during training.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

3 participants