Open felixdae opened 1 week ago
https://github.com/XLabs-AI/x-flux/blob/47495425dbed499be1e8e5a6e52628b07349cba2/train_flux_deepspeed_controlnet.py#L219
as this line show, why we use a Logit-normal distribution for t here. from the flow matching paper https://arxiv.org/abs/2210.02747, t is sampled from uniform distribution U(0,1)
https://github.com/XLabs-AI/x-flux/blob/47495425dbed499be1e8e5a6e52628b07349cba2/train_flux_deepspeed_controlnet.py#L219
as this line show, why we use a Logit-normal distribution for t here. from the flow matching paper https://arxiv.org/abs/2210.02747, t is sampled from uniform distribution U(0,1)