unslothai / unsloth

Finetune Llama 3.2, Mistral, Phi & Gemma LLMs 2-5x faster with 80% less memory
https://unsloth.ai
Apache License 2.0
16.49k stars 1.14k forks source link

请加入gemma2 #707

Open win4r opened 3 months ago

win4r commented 3 months ago

可以加入这个gemma2的微调吗 https://youtu.be/N8yBqEZvvsU

danielhanchen commented 3 months ago

Working on it!

Qualzz commented 3 months ago

excited for that one !

danielhanchen commented 3 months ago

Finally got it to work! Try updating Unsloth via (if you installed Unsloth locally - Colab and Kaggle just need to refresh)

pip uninstall unsloth -y
pip install --upgrade --force-reinstall --no-cache-dir git+https://github.com/unslothai/unsloth.git

We also have a Colab notebook for Gemma 2: https://colab.research.google.com/drive/1vIrqH5uYDQwsJ4-OO3DErvuv4pBgVwk4?usp=sharing