allenai / ms2 Goto Github PK

This dataset looks awesome! Will this be available in the HuggingFace Dataset library soon?

Hi，I use huggingface env to load LED model for getting embeding vector， but the error shows that dim size is fault about [16389,768] and [1072, 768].

However, I have Seen, in this repo, you used LED model and import transformers, So could you tell me what I should pay attention to and how to make it work.

if you see, please reply me soon,Thanks!

emnlp poster

Hi！I use your code to achieve text2text task and use the BART configuration, but in the training process

• gradient overflow always occurs in every epoch
• I find all epoches (except the last one) cannot process all the steps, skipping the rest steps at every epoch

If you see, please reply me soon,Thanks!

Hi！I use your code to achieve text2text task and use the BART configuration, but in the training process

• gradient overflow always occurs in every epoch
• I find all epoches (except the last one) cannot process all the steps, skipping the rest steps at every epoch

If you see, please reply me soon,Thanks!

Excuse me, the folder path of MEDLINE_ABSTRACTS_PATH.MEDLINE_ABSTRACTS_PATH in codes/data_preparation/Statistics_Named_Entities_PubMed_and_ICD_Summ.py is not found or generated from other .ipynb.
Is there any way I can find them or am I missing something?

May I have your email address if I have further question?