SberQuAD is one of the basic reading comprehension datasets for Russian language. However after Sberbank Data Science Journey 2017 materials were not available officially in open source. The quality of the dataset have also been criticized a lot for example in the paper of the same name. Now, we publish an original sberQuAD dataset with a number of improvements:
SberQuAD (like in origin competition) evaluation employs two metrics to assess model performance:
The competition is realized in the endless mode. We provide data for training and offer a private testset. You can score your solution and make public submit on the leaderboard. We don't require obligatory the systems code, but you are free to share your results and systems. This dataset is distributed under the CC BY-SA 4.0 license.
Start: Feb. 9, 2021, midnight
Never
You must be logged in to participate in competitions.
Sign In# | Username | Score |
---|---|---|
1 | antivist | 0.87 |
2 | tim_atnashev | 0.87 |
3 | pokld | 0.87 |