File size: 311 Bytes
83839c5
 
f6b382f
 
83839c5
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
The Reader model is for Korean Question Answering

The backbone model is deepset/xlm-roberta-large-squad2.

It is a finetuned model with KorQuAD-v1 dataset.


As a result of verification using KorQuAD evaluation dataset, it showed approximately 87% and 92% respectively for the EM score and F1 score.

Thank you