| `best_model_pretrain20.pth` and `best_model_pretrain10.pth` are obtained after pretraining for 20 and 10 epochs, respectively, on a dataset containing ten million samples. |
| `best_model_pretrain20.pth` and `best_model_pretrain10.pth` are obtained after pretraining for 20 and 10 epochs, respectively, on a dataset containing ten million samples. |