Open rivaldophilip opened 5 months ago
More details using faulthandler
It seems the fault happens at the first time of NN computing in stage 2. Could you try to feed forward the panel conditioner in ranni/panel_conditioner.py
with input control = (torch.rand(1, 2, 3, 96, 96).cuda(), torch.rand(1, 2, 1024).cuda())
?
Segmentation fault error when doing panel-to-image
Error also happens on sample images / generation