Ming Zhong
Ming Zhong
We truncate each document to 512 tokens and feed them to MatchSum because the pre-trained models (BERT, RoBERTa) has a maximum length limit. One possible solution is to train the...
I re-test the ROUGE score and still get the same result as I reported. I'm not sure what caused this discrepancy. If you can't get the same results on other...
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](https://github.com/maszhongming/UniEval/blob/main/pseudo_data_summ.py)) for your reference.
Hi, the data for the consistency dimension is from [another paper](https://github.com/zide05/AdvFact).
Sorry, but our current version of UniEval only supports English.
Sorry for the annotations that may be problematic! In fact, @WadeYin9712 and I are responsible for the review of Product domain and (part of) Committee domain in QMSum dataset. It...
Hi, we have updated the annotations of Bmr006 and Bed008. We will continue to look for problematic annotations and fix them.
Hi, thanks for your interest in our work! Unfortunately I don't have the Locator code and pre-trained model right now, but I have released the output of our Locator. You...
I’d like to clarify a few points regarding your questions: 1. The ROUGE metric correlations are from the BARTScore paper, but I believe all ROUGE scores are calculated against annotated...
Same issue here. Happy to help test if a fix is in progress.