Why is the load_from attribute in the train config for snan-print set to a pre-trained SCM model?

[PeiqinSun]

Hi, your open-source work is excellent. While reproducing snan-print, I noticed that in the training script, the load_from attribute for the student is set to a checkpoint that can already produce good results in one step. Is this the intended behavior? Shouldn't the student be initialized with the teacher instead? This seems to be inconsistent with the description in the paper. If I want to understand your training process better, should I change it to initialize the student with the teacher?
Looking forward to your response.
Config: https://github.com/NVlabs/Sana/blob/main/configs/sana_sprint_config/1024ms/SanaSprint_1600M_1024px_allqknorm_bf16_scm_ladd.yaml#L23