The animated graphics were chosen from the periods when the performance of the model was changing drastically. Afterwards, a summary of the training is available.
| epoch | 1 | 3 | 125 |
| results |  |
 |
 |
| epoch | 165 | 185 | 219 |
| results |  |
 |
 |
 |
| epoch | 1 | 2 | 12 |
| results |  |
 |
 |
| epoch | 27 | 45 | 413 |
| results |  |
 |
 |
 |