Skip to content

Latest commit

 

History

History
24 lines (17 loc) · 625 Bytes

File metadata and controls

24 lines (17 loc) · 625 Bytes

Development Notes

Setup

conda create -n sqlft python=3.10 -y
conda activate sqlft
pip install -r requirements.txt

Workflow

  1. Edit config.yaml for your experiment
  2. Train: python src/fine_tune.py
  3. Evaluate: python src/evaluate.py
  4. Inference: python src/inference.py

The training script splits the dataset into train/validation sets automatically. The validation set is saved to outputs/val_dataset/ so that evaluate.py can use the same held-out data.

Contribution Tips

  • Keep PRs focused and small
  • Explain your reasoning in the PR description
  • Use meaningful commit messages