Full-KL co-distillation
Named by 2 kept items in this workspace.
Search every workspace for Full-KL co-distillation. It matches the name as written, so it can find other things that share it.
The items that name it 2
Meta used it from Muse Spark in-training logits for Muse Glimmer, then cross-entropy for the final 5%.
All techniques in this workspace.