I've noticed some discrepancies between the LTX 2.3 inference dubit functionality and the original paper and repository, such as the mask mechanism, the order of the original and reference videos, and the use of the distillation model. I'd like to ask if it's possible to obtain details about the training process in the new version. Thank you so much！
I've noticed some discrepancies between the LTX 2.3 inference dubit functionality and the original paper and repository, such as the mask mechanism, the order of the original and reference videos, and the use of the distillation model. I'd like to ask if it's possible to obtain details about the training process in the new version. Thank you so much！