xcfeng

Hi, thanks for this great integration of NAT codes. Could you please provide an example to show how to train a `GLAT` model?

Dear Robin, thanks for your help, I have successfully finished the training process, could you kindly provide the test script?

Sincerely thanks！

update: working on this issue~

For the second problem **Data pre-processing**: 1. If you just want to generate a meeting summary without calculating ROUGE scores, you **do not** need a `test.tgt` file, since this file...

感谢关注！ 1. 正常，AMI数据集平均输入词语数量为4000+，需要时间较久。 2. 采用格式为我公开在Google Drive格式，没有其他特殊格式。可以尝试处理AMI或者SAMSum是否成功。 ``` [["Hannah needs Betty's number but Amanda doesn't have it . She needs to contact Larry .", "Hannah : Hey , do you have...

Hi, we use [F-score](https://github.com/xcfcode/PLM_annotator/blob/main/bart/py_rouge_test.py) for ROUGE. For baseline data, I upload them to [here](https://drive.google.com/drive/folders/1wLea1LdEv1jFQMtXr3bXJFiKvZnrGQLO), under the `clean_samsum` dir.

Hi, `py-rouge` does lowercase inside the pkg: https://github.com/Diego999/py-rouge/blob/16f225b1f46e9d382f1bf5170546da218ee98003/rouge/rouge.py#L697 and I provide the test code, have you tried this one? https://github.com/xcfcode/PLM_annotator/blob/main/bart/py_rouge_test.py

Thanks for your information, how do you get the above results, could you please show me some details, the rouge-l score looks much higher.

On my side, I just run `python py_rouge_test.py -c summaries/samsum.txt `