| # sbc | |
| ## 0. Setup environment | |
| run `install.sh` to setup `unsloth` package | |
| ## 1. create a config file | |
| check config/ folder for examples | |
| ## 2. Data format | |
| make sure your data is in `https://huggingface.co/datasets/HuggingFaceH4/ultrafeedback_binarized` format | |
| ## 3. train the model | |
| train the model by command `python3 run.py configs/Qlora_config.yaml` |