Skip to content

training data to finetune #339

@abpani1994

Description

@abpani1994

I was referring to the synthetic data generation example.
https://github.com/urchade/GLiNER/blob/main/examples/synthetic_data_generation.ipynb

THe code creates dat without any index spans. Will it work with gliner fine tuning as gliner supports span based entities .
Also what is the optimal way to generate data. Is it token level spans or character level spans.
Can I combine both data types in one fine tuning step?

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions