Chen Liang

We optimize the student model only based on the losses of the tokens which the student model predicts with high confidence. The corresponding line in the code is https://github.com/cliang1453/BOND/blob/32f26988a58ee44eb4f50772c6d6c6eb116c83cf/model_utils.py#L88.

What you understand is correct. I was saying that "we optimize the student model only based on the losses of the tokens which the teacher model predicts with high confidence"....

Hi @zwhe99 , I have no upcoming plan of releasing it - I no longer have access to the server that stores the code, unfortunately. However, it should be easy...

> I have submitted a fix in TFDS. Could you retry with tfds-nightly? If it's working, we'll release a new version of TFDS. It seems that using ```tfds-nightly 4.8.3.dev202303300044``` gives...

Hi Chengfei, thanks for your interests in TED. We will be working on adding the GLUE codes and hyperparameters soon. Please stay tuned.

Hi @jian53286 , the code for GLUE has been released. Hi @aaronma2020 , we will work on adding GPT-2 soon.