Is it safe to delete the older models, such as G_0 and D_0, once the training is finished?

[Mixeykusje]

Is it safe to delete the older models, such as G_0 and D_0, once the training is finished? I currently have around four G and D models saved due to a lower eval interval. In my case, I only utilize the latest G model, specifically G_601, and I would like to remove the others to free up storage space.

Should I delete the older models only after completing the training, or is it also acceptable to delete them while temporarily suspending the training process? I understand that these models serve as checkpoints, allowing me to resume training from a specific number of epochs onwards. My intention is to optimize hard disk space usage without compromising the model's quality.

[seoeaa]

Make a backup and check