real-stanford / diffusion_policy

[RSS 2023] Diffusion Policy Visuomotor Policy Learning via Action Diffusion
https://diffusion-policy.cs.columbia.edu/
MIT License
1.09k stars 203 forks source link

dataset format #89

Open deavn2236 opened 3 weeks ago

deavn2236 commented 3 weeks ago

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