Updated model dev HPT to save the model for each trial then register the best model

informatics-lab / ukrse_2022_mlops_walkthrough

Material for UK RSE 2022 Conference Walkthrough on MLOps for RSEs

15 stars 1 forks source link

Updated model dev HPT to save the model for each trial then register the best model #16

Open stevehadd opened 2 years ago

stevehadd commented 2 years ago

Currently ray tune doesn't inherently save each model. After doing hyperparameter tuning we want to look for which model gave us the best results and be able to save or ideally register that model in our ML Flow artifact repository