Fine-tuning Llama 3 on Wikipedia Datasets for Low-Resource Languages

Fine-tuning Llama 3 on Wikipedia Datasets for Low-Resource Languages

Trelis Research via YouTube Direct link

Inspection of losses and results

8 of 10

8 of 10

Inspection of losses and results

Class Central Classrooms beta

YouTube videos curated by Class Central.

Classroom Contents

Fine-tuning Llama 3 on Wikipedia Datasets for Low-Resource Languages

Automatically move to the next video in the Classroom when playback concludes

  1. 1 Fine-tuning Llama 3 for a low resource language
  2. 2 Overview of Wikipedia Dataset and Loss Curves
  3. 3 Video overview
  4. 4 HuggingFace Dataset creation with WikiExtractor
  5. 5 Llama 3 fine-tuning setup, incl. LoRA
  6. 6 Dataset blending to avoid catastrophic forgetting
  7. 7 Trainer setup and parameter selection
  8. 8 Inspection of losses and results
  9. 9 Learning Rates and Annealing
  10. 10 Further tips and improvements

Never Stop Learning.

Get personalized course recommendations, track subjects and courses with reminders, and more.

Someone learning on their laptop while sitting on the floor.