[TODO]: Add NER from Thai-NNER dataset

[wannaphong]

From Thai-NNER, The dataset has 4,894 docs.

The dataset is licensed under CC-BY-SA 3.0.

dev.txt
train.txt

[wannaphong]

@pavaris-pm @MpolaarbearM Do you interested to train the model?

[MpolaarbearM]

Yes I could try training NER with the new model

[pavaris-pm]

@pavaris-pm @MpolaarbearM Do you interested to train the model?

Sure. Maybe I'll made an experiment with different models architecture and made PR then

[MpolaarbearM]

I completed the training on the NER model and performed some fine-tuning, but the outcomes were not meeting expectations. I plan to make adjustments to the model in an attempt to enhance its performance. https://huggingface.co/lunarlist/NER_thai_phayathai

[wannaphong]

I completed the training on the NER model and performed some fine-tuning, but the outcomes were not meeting expectations. I plan to make adjustments to the model in an attempt to enhance its performance. https://huggingface.co/lunarlist/NER_thai_phayathai

The dataset is very hard. I think you can turn some parameter.

[wannaphong]

I rethink about this issue. You can actually use Thai-NNER and get just the top tag of nner. https://github.com/vistec-AI/Thai-NNER/blob/main/thai_nner/__init__.py#L31

Don't need to train any model!

[MpolaarbearM]

Okay, I'll look into this krub