Pause and resume training question

[smlkdev]

I would like to pause training and resume it later. Which files should I keep?

Is it enough to keep only the following files:

(*.bert.pt, *spec.pt, *.wav) for each audio file in the dataset, along with config.json, val.list, train.list, metadata.list.cleaned, and metadata.list?

Are these files sufficient to resume training later?

Do I need any files from ~MeloTTS/melo/logs/data/ to continue training, such as train.log, G_XXXXX.pth, DUR_XXXXX.pth, D_XXXXX.pth, config.json, or are these files only necessary when using my trained voice (specifically config.json and G_XXXXX.pth so I can use infer.py script)?

[KiON-GiON]

Good question. Since I'm currently doing a finetune using cloud services, I need to know this as well to save compute.