Repository navigation
How is the performance if we train with sCM loss only? #111
Description
Hi @zhengkw18,
In the paper, you compared the rCM with DMD loss, I am wondering what if you train the model with sCM loss only? Do I need to turn the hyperparameters a lot if I want to use sCM loss only?
Btw, would you mind sharing your loss curves/grad norms over training time?
Thanks in advance!
Meng
Activity
zhengkw18 commentedon Feb 5, 2026
Training with sCM only will result in high generation diversity, but the quality is low. There are sCM-only results in the paper and the rCM repo.
Loss curves and grad norms are not important in sCM or DMD distillation, unless you encounter bad results or potential bugs.