qbara_qahira
February 27, 2025 ยท View on GitHub
To be implemented! Coming soon!
Installation
To load models in 4bits with transformers and bitsandbytes, you have to install accelerate and transformers from source and make sure you have the latest version of the bitsandbytes library. After installing PyTorch, you can achieve the above with the following command:
pip install -U -r requirements.txt
Fine-tune
-
Change the
bnb.pyin your peft path(python path/peft/tuners/lora/bnb.py) with the new one. -
Change the
layer.pyin your peft path(python path/peft/tuners/lora/layer.py) with the new one.
Merge
Use the merge.py to merge the base_model and the adapter
Acknowledgements
Our code is based on QLoRA