Replies: 1 comment 1 reply
-
@ignaciot I would recommend using the GLiNER fine-tuning branch instead of the fine-tuning jupyter notebook. https://github.com/urchade/GLiNER/tree/training This branch was easy to follow given the steps listed. Note that it uses GLiNER 0.2.2, which is both behind and ahead of the main branch, so I recommend setting up a private environment. Look at the pilener.json file for the format the training data needs to be in. |
Beta Was this translation helpful? Give feedback.
1 reply
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
Hi there, thanks for providing such a straightforward finetuning notebook. I'm experimenting with fine-tuning GLiNER and I was wondering, is the
data.json
file in the finetuning example provided in any specific standard format? (and is there a tool that readily outputs that specific JSON format?)Asking before writing a custom CONLL/whatever converter to that expected JSON structure...
Beta Was this translation helpful? Give feedback.
All reactions