WANDB + K-FOLD + Multi-GPU (SLURM)

[YerePhy]

Questions

• Is it possible to train a model in a K-Fold set up with the Loop API and use WANDB as logger?
• How are the logs managed? What is logged in WANDB? How WANDB shows the results of the different folds?
• Is it possible to do this training with Multiple GPU on a SLURM cluster?
• What about NVIDIA DALI? Can it be finally used it with the set up described above?