“28e22a6039b001a49aaa19a54b46fd66946e7ee9”上不存在“...mgmt/node_mgmt/git@gitcode.net:taosdata/tdengine.git”
  • E
    Fix sparse training for trainer_count=1 (#204) · 28bc05b1
    emailweixu 提交于
    * Fix sparse training for trainer_count=1
    
    For trainer_count=1, the gradient machine is NeuralNetwork, which does not create parameter buf for PARAMETER_GRADIENT for sparse update in Parameter::enableType. But gradient parameter buf is still used in SgdThreadUpdater.
    
    * Minor update to comment
    28bc05b1
MultiGradientMachine.cpp 25.6 KB