Regarding the format of the DPR dataset

[mc112611]

I have encountered a little trouble and need help.

I want to train a DPR model based on my own dataset

But I am not very clear about the format of the dataset. Could you please explain 'positive'_ ctxs‘，‘negative_ ctxs‘，‘hard_ negative_ ctxs‘。

What is the abbreviation for ctxs.

I don't know if the score below 'text' represents the score of the answer, and I also don't know the meaning of 'title_score'.

And the relationship between title, text, score, and title score

Thanks again

[mc112611]

I don't seem to have read the tutorial well. There are detailed explanations in the tutorial, and I apologize for my carelessness.