projects / platform / core / ml / nntrainer.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: 58fd70e) | patch
[dynamic-training] Add dynamic training using derivatives
authorParichay Kapoor <pk.kapoor@samsung.com>
Tue, 5 Jan 2021 15:16:03 +0000 (00:16 +0900)
committerJijoong Moon <jijoong.moon@samsung.com>
Mon, 25 Jan 2021 12:58:59 +0000 (21:58 +0900)
commita54ef34d1ec7ce57f3aac9414d56040e153bcc73
tree1807735a62a889c07f7dffca9cb789c01fd0c2e5tree | snapshot
parent58fd70e735d2ad7ad8e390726dbb06de3c77c6a5commit | diff
[dynamic-training] Add dynamic training using derivatives

Added dynamic training using derivatives where the decision to
apply the gradient is calculated using the derivative received
without calculating the gradient itself.

Signed-off-by: Parichay Kapoor <pk.kapoor@samsung.com>
jni/Android.mk diff | blob | history
nntrainer/models/dynamic_training_optimization.cpp [new file with mode: 0644] blob
nntrainer/models/dynamic_training_optimization.h diff | blob | history
nntrainer/models/meson.build diff | blob | history
nntrainer/models/neuralnet.cpp diff | blob | history
nntrainer/models/neuralnet.h diff | blob | history
nntrainer/tensor/blas_interface.cpp diff | blob | history
nntrainer/tensor/blas_interface.h diff | blob | history
nntrainer/tensor/tensor.cpp diff | blob | history
nntrainer/tensor/var_grad.h diff | blob | history
Domain: Machine Learning / ML Framework;