• Xuanlei Zhao's avatar
    [moe] support optimizer checkpoint (#5015) · f71e63b0
    Xuanlei Zhao authored
    * Refactor MoE Manager setup method
    
    * unshard optim ckpt
    
    * optim io
    
    * update transformer version
    
    * update requirements
    
    * update ckpt
    
    * update ckpt
    
    * update ckpt
    
    * fix engine
    
    * fix engine
    f71e63b0
test_grad_handler.py 2.63 KB