Replies: 1 comment
-
|
yes, my guess is that the segmentation is not optimal for how we label tokens during training. See the label is put on for example "Al+" although it should be on the word itself. |
Beta Was this translation helpful? Give feedback.
0 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
Hello
My task like NER, I have tried different arabert versions (with Farasa and without Farasa), the results without Farasa are much better. Is there an explanation for that?
Thanks
Beta Was this translation helpful? Give feedback.
All reactions