the training iteration is abnormally large

[bolin-chen]

I used 4 gpus(2080ti) to train swin_small with config swin_small_patch244_window877_kinetics400_1k.py. The dataset I used is hacs(50w videos). The following is some of my training log:

I find that the train iteration is abnormally large based on my config(dataset size(50w) and batch_size(8)), which leads to the long training time. Is that normal?

[MartinXM]

Check your optimizer_config, set update_interval=1 if your GPU memory is large enough, it would make your training faster.

[bolin-chen]

Get it. It seems that the actual batch size is batch_size / update_interval.