search

maszhongming / UniEval

Repository for EMNLP 2022 Paper: Towards a Unified Multi-Dimensional Evaluator for Text Generation
MIT License
184 stars 25 forks source link

How to finetune UniEval? #5

Open felipemaiapolo opened 1 year ago

felipemaiapolo commented 1 year ago

Dear authors,

Thank you for the great work. I wanted to finetune UniEval on other datasets, let's say XSum (instead of CNN/DailyMail). What do you think the best way is to proceed? Would you happen to have the code to generate the pseudo data? I could not find it. What do you think you could advise?

Thank you! Felipe

maszhongming commented 1 year ago

I apologize for not having the time to structure the code for pseudo-data construction recently. However, I've uploaded an initial version (pseudo_data_summ.py) for your reference.

felipemaiapolo commented 1 year ago

Thank you for sharing your code. I will review it soon and let you know if I have more questions!

felipemaiapolo commented 1 year ago

Dear authors,

I realized that the function which generates non-consistent summaries is missing in (pseudo_data_summ.py). Is this code available anywhere else? Thank you!

maszhongming commented 1 year ago

Hi, the data for the consistency dimension is from another paper.