Authors: Kelvin Guu*, Tatsunori B. Hashimoto*, Yonatan Oren, Percy Liang (* equal contribution)
- A detailed description of the training algorithm is now available here.
- We are drafting a more detailed README in the README branch (see here for dataset links)
- This is research code meant to serve as a reference implementation. We do not recommend heavily extending or modifying this codebase for other purposes.
If you have questions, please email Kelvin at