master
optimize arm-package size
gpu demo
clean code
[MS][LITE] fix loss_name_ api string with cxx11-abi=0
add sub communication group for distributed inference
ckpt_support_bf16
modify foc
refine the delegate api
optimize tfrecord dataset
Move some basic type defines to mindapi
remove lite expression
add MSKernel and BaseKernel
merger TTS PR from br_base to master
730需求合入主线
RunnerConfig add Set/Get DeviceIdList api
Train support export reasoning-model's weights for micro
add python error code and refactor c++ status code
device data
adapt dllexport for windows