leesureman / sequence-labeling-early-exit Goto Github PK

Hi @LeeSureman ,

I am a bit confused with the modeling_bert_ee code.

If the training mode is 'joint_sample_copy' as in further_train stage:

The loss will computed in the if branch of line 2101:

if self.args.train_mode == 'joint_sample_copy_label' or (self.args.train_mode == 'joint_sample_copy' and self.args.copy_supervised == 0):

and then recomputed again and overwritten in the if branch of line 2190:

elif 'joint' in self.args.train_mode:

Then, the if branch of line 2101 will be useless.

May I know why is that?

Thank you!

Hi, I can't find requirements.txt file

Can you please tell me what the dependencies are?

Thank you in advance!

Coule I know when the code will come?

请问用已训练好的模型，新的sentences输入的时候，怎么来进行预测呢

An error happened in load_data_.py, line 231, 'yangjie_rich_pretrain_bigram_path' is not defined. But in paths.py, this variable is not mentioned. And I also want to know which dataset in the paths.py should be used first for testing the programme.

看main.py的時候有些导入的包是找不到的，所以什么时候更新呢？