Two teammates trained the same network on the same data for 4,000 steps. Each picked one fixed learning rate and kept it for the whole run, but neither wrote down which. All you have is the training loss, measured on the full training set at a few checkpoints:
| step | 0 | 250 | 500 | 1,000 | 2,000 | 3,000 | 4,000 |
|---|---|---|---|---|---|---|---|
| Run P | 2.30 | 2.19 | 2.08 | 1.89 | 1.57 | 1.30 | 1.07 |
| Run Q | 2.30 | 0.96 | 1.18 | 0.94 | 1.13 | 0.98 | 1.05 |
Both runs finish at almost the same loss.
What should each teammate do to their learning rate before the next run?
Select all that apply.