Inference batch size

[inakierregueab]

As far as I could understand the code, if N slices are generated, the inference is performed N times on batch 1. Are you going to implemente the customisation of this batch size? Ideally a single forward pass with batch size being N, the number of slices. That would speed up inference!:) ty