[llama4][auxiliary-loss-free load balancing] update expert_bias without backward hooks#1304
Open
hann-wang wants to merge 2 commits intopytorch:mainfrom
Open
[llama4][auxiliary-loss-free load balancing] update expert_bias without backward hooks#1304hann-wang wants to merge 2 commits intopytorch:mainfrom
hann-wang wants to merge 2 commits intopytorch:mainfrom