Are you planning to release the source code that you used to train sodaverse?

skywalker023 / sodaverse

🥤🧑🏻‍🚀Code and dataset for our EMNLP 2023 paper - "SODA: Million-scale Dialogue Distillation with Social Commonsense Contextualization"

https://aclanthology.org/2023.emnlp-main.799/

MIT License

221 stars 13 forks source link

Closed justlearntutors closed 1 year ago

justlearntutors commented 1 year ago

I am looking to add more data to the model and other modifications.

skywalker023 commented 1 year ago

Currently, we do not have detailed plans on releasing the training code, but we used the code here: https://github.com/google-research/t5x ☺️