search

unslothai / unsloth

Finetune Llama 3.1, Mistral, Phi & Gemma LLMs 2-5x faster with 80% less memory
https://unsloth.ai
Apache License 2.0
15.11k stars 1.01k forks source link

Epoch Number #605

Open nichellehouston opened 3 months ago

nichellehouston commented 3 months ago

For fine tune llama-3 to language translation what is best epoch number and how set in: https://colab.research.google.com/drive/135ced7oHytdxu3N2DNe1Z0kqjyYIkDXp?usp=sharing

danielhanchen commented 2 months ago

1 to 5 - I would say 3