T2TTrainer.cpp
18.6 KB
-
Support 0d tensor(scalar) for ReduceSumAll function. · 84ea960aUse tensor as the bottom implementation and support plain scalar. Note: DO NOT support the backward for ReduceSumAll function.
liyinqiao committed
Use tensor as the bottom implementation and support plain scalar. Note: DO NOT support the backward for ReduceSumAll function.