No difference between RougeL and RougeLsum

[FeiWang96]

🐛 Bug

To Reproduce

I finetune BART on CNN/DailyMail dataset using lightning transformers.

Code sample

https://github.com/PyTorchLightning/lightning-transformers/blob/master/lightning_transformers/task/nlp/summarization/model.py#L43

Expected behavior

Each document and summary in CNN/DailyMail have multiple sentences, so RougeL and RougeLsum should be different.

Environment

• OS (e.g., Linux): Linux
• Python & PyTorch Version (e.g., 1.0): Python 3.9, Pytorch 1.10
• How you installed PyTorch (conda, pip, build command if you used source): pip
• Any other relevant information:

Additional context

I've checked the code for computing RougeL and RougeLsum.
It seems that the only difference is that a separator '\n' is inserted after each sentence when calculating RougeLsum.
However, the '\n' is not treated as a special token when calculating lcs and the final score.

[github-actions]

Hi! thanks for your contribution!, great first issue!

[SkafteNicki]

cc: @stancld