❓ Questions/Help/Support

Hi @vfdev-5 ,

We have some requirements for very big dataset:
See we have 100 iterations for 1 epoch, we want to call run() to train on the first 10 iterations data, do some other things for the model and then call run() to train on the second 10 iterations data, ...
Is it possible in ignite now?
I found the the iter for dataloader in ignite is always from beginning?
https://github.com/pytorch/ignite/blob/v0.4.2/ignite/engine/engine.py#L771

Thanks.