Sewon Min

Sorry for the inconvenience. Could you change [this line](https://github.com/seominjoon/qrn/blob/master/dialog/model.py#L161) `6 : range(11)` to be `6: range(8)` ?

Yes! We don't have any unanswerable questions. If we find some questions unanswerable during annotation, we excluded them in the dataset.

Hi, thanks for your interest! I do not have the checkpoint and prediction right now, but I'll be able to train the model again and make it public. Unfortunately I...

Hi, I'm sorry for the delay. Here is the model checkpoint for NarrativeQA: [link](https://drive.google.com/file/d/1F_LW73hVMiWaVs2LpnFcQIAqfMatJose/view?usp=sharing)

Hi @DbrRoxane, for L315, it actually won't matter because the answer mask is 1. For L281, it should be "3" as it means there is no answer ("1" and "2"...

Oh I got it. You are correct. It should be 0 or be changed into `switch!=3`. Thanks for catching! It shouldn't affect the performance though and will only cause the...

감사합니다. 말씀하신 대로 하여 해결했습니다. 도움이 정말 많이 되었네요. 올려주신 CBP 덕분에 MCB for Visual QA를 구현했는데 혹시 나중에 논문 전체를 구현하게 된다면 참고해보세요. https://github.com/shmsw25/mcb-model-for-vqa

Hi @laiviet, thanks for your interest with our work!! In this case, I think you need to modify some code in [factscorer.py](https://github.com/shmsw25/FActScore/blob/main/factscore/factscorer.py) - but it shouldn't be too complicated! What...

Hi @caiqizh, thank you for your interest in our work. Here is the prompt we used for ChatGPT:  Here are two hyperparameters: * temp=0.7 for both ChatGPT and InstructGPT...