Dear @ZINDI,

The timing limitation appears too generous for inference than for model training

"T4 GPU, maximum 9h training, maximum 3h inference"

Can we use all 12 hours for training and inference without any 9-hour specific timing for training?

Thanks