projects / platform / core / ml / nntrainer.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: f2b0d35) | patch
[layer] Bug fix for softmax operation
authorParichay Kapoor <pk.kapoor@samsung.com>
Tue, 5 Oct 2021 05:25:29 +0000 (14:25 +0900)
committerJijoong Moon <jijoong.moon@samsung.com>
Fri, 8 Oct 2021 06:30:06 +0000 (15:30 +0900)
commit07f0dc34d11c43b0abab9a3c92ef5654e6288d11
tree351a30c6743cbd66f512a829210bac645b9d0db8tree | snapshot
parentf2b0d35ed292bdd3239c4ff10bd7d9500f953dabcommit | diff
[layer] Bug fix for softmax operation

Current implementation of softmax operation applies flatten on the
tensor unintentionally and calculates softmax on the last 3 dimensions
of the given tensor.
This patch updates the softmax operation to apply its operations on the
last dimension only.

Signed-off-by: Parichay Kapoor <pk.kapoor@samsung.com>
nntrainer/layers/acti_func.cpp diff | blob | history
nntrainer/tensor/tensor.h diff | blob | history
Domain: Machine Learning / ML Framework;