diffusion_policy icon indicating copy to clipboard operation
diffusion_policy copied to clipboard

Published 20 hours ago • verified publisher icon real-stanford

dataset format

Open deavn2236 opened this issue 8 months ago • 0 comments

Why is it that when training dp on the robomic dataset, hdf5 is not directly used but zarr is used instead? Isn't the speed of hdf5 much faster than zarr

deavn2236 avatar Jun 13 '24 07:06 deavn2236