add back the topics with empty neg docids so that the train info is full

remove empty neg doc from finnish training set

Merge branch 'main' of https://huggingface.co/datasets/castorini/mr-tydi

add text and title to dev and test split

Update README.md

add script to load dataset

add readme

add pos doc id in dev and test

arabic train; combined train

init

initial commit