search

Alibaba-NLP / ACE

[ACL-IJCNLP 2021] Automated Concatenation of Embeddings for Structured Prediction
Other
296 stars 44 forks source link

Implementation of Abstract Extraction #18

Closed ntdas closed 2 years ago

ntdas commented 2 years ago

Thank you for your awesome work, but after looking at part of the code I realized the training code for abstract Extraction is currently under "rejected list". Could you explain a little bit how you obtained the results for SentEval dataset? I am trying to reproduce the results and hopefully use them for my own project.

Sincerely.

wangxinyu0922 commented 2 years ago

Hi,

I'm not sure the meaning of the "rejected list". We take the task as a sequence labeling task and format the datasets into BIO formats. I think you may need the aspect extraction datasets and I uploaded it on Onedrive. You can follow the guide in README to train your model.

ntdas commented 2 years ago

Thanks for clarifying, I will look into it more carefully.