v0.2.0
Highlights since v0.1.0: - Multi-duration support: refactor of tasks, training, and eval (#38) - Eval suite improvements + select_model (#37) - Sample task generator - GCS sync and model upload scripts (#42) - Lower minimum Python from 3.12 to 3.11 (#50) - Numerous bug fixes and test additions