There were a lot of request from friends to draft this, I won't really call it a starter notebook, but it's a very simple approach for anyone to understand. I've also taken out quality time to explain the code cells as high level as I can. This single notebook should get you to ~0.25.
I'm hoping hackers will be able to ensemble with their current solution to make the jump, and hopefully learn something new. Please feel free to send GitHub merge requests for updates or contributions.
Colab link: Swahili News Classification FastAI xBlurr xHugginface
GitHub link: GitHub Swahili-News-Classification-Challenge
Keep Learning! Keep Winning!
Thanks so much for sharing.
@Femi (a.k.a Professor) You and this your fastai,,,, I will love to see you working for Fastai one day. Smiles,,,,,😁. Well done.....👍. Thanks to "mbeukman" for this model,
pretrained_model_name = "mbeukman/xlm-roberta-base-finetuned-swahili-finetuned-ner-swahili"
😅, thanks a lot too @MICADEE.
😁👍