Why do you use batch size = 1?

[coolbay]

Very good work and neat code. I have the following question regarding the batch size.

Could you please explain the reason that you use batch size 1? Usually for other tasks, a larger batch size is perferred to fill the entire GPU. With batch size = 1, you only use a small portion of even an 11GB GPU. So you don't use a larger number for the performance purpose? Why does batch size=1 fit the task of video compression?

Also I notice that in HNeRV, batch size = 2 is used, which is also a quite small number. What will happen if you use a larger batch size to fill in the GPU?

Thank you very much! Hope to hear from you soon.