Prepare custom data for to train DPR model

[vibha0411]

I am trying to prepare custom data to train DPR model.
But I dont understand what the score, title score refer to:
eg:
"positive_ctxs": [
{
"title": "Does He Love You",
"text": "Does He Love You "Does He Love You" is a song written by Sandy Knox and Billy Stritch, and recorded as a duet by American country music artists Reba McEntire and Linda Davis. It was released in August 1993 as the first single from Reba's album "Greatest Hits Volume Two". It is one of country music's several songs about a love triangle. "Does He Love You" was written in 1982 by Billy Stritch. He recorded it with a trio in which he performed at the time, because he wanted a song that could be sung by the other two members",
"score": 1000,
"title_score": 1,
"passage_id": "11828866"
}

Does anyone have any ideas on this ?

[JasperLS]

This question was also asked here with regard to DPR models. It's related to the BM25 retriever. I also noticed the title_score is 1 only one time per query, while score seems to reflect a continuum from 0 to 1000 depending on relevance to the query.

I don't think you need these scores for training on custom dataset. You might wanna try only setting positive context / hard negative context first.