You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
is there any issue here, if i want to train the model for num_frames less than 17, the transformed timestep will always be zero irrespective of the original timestep because ratio_time will always be zero?
The text was updated successfully, but these errors were encountered:
https://github.com/hpcaitech/Open-Sora/blob/786d4e7afcb2bb05da47539cdf23e04520664741/opensora/schedulers/rf/rectified_flow.py#L28C5-L36C1
is there any issue here, if i want to train the model for num_frames less than 17, the transformed timestep will always be zero irrespective of the original timestep because ratio_time will always be zero?
The text was updated successfully, but these errors were encountered: