Description
To avoid unnecessary retraining with the same configuration, the tuning script should save the best-performing model during the tuning process, particularly when no additional training steps are required.
Implementation
-
After each trial t:
- If
model_t performs better than best_model_so_far:
- Update
best_model_so_far = model_t
-
At the end of tuning:
Description
To avoid unnecessary retraining with the same configuration, the tuning script should save the best-performing model during the tuning process, particularly when no additional training steps are required.
Implementation
After each trial t:
model_tperforms better thanbest_model_so_far:best_model_so_far = model_tAt the end of tuning:
best_model_so_far