projects / platform / core / ml / nntrainer.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: 59a4c4a) | patch
[rnncell] implement multiout for rnncell
authorhyeonseok lee <hs89.lee@samsung.com>
Tue, 26 Apr 2022 09:57:21 +0000 (18:57 +0900)
committerJijoong Moon <jijoong.moon@samsung.com>
Wed, 11 May 2022 04:52:11 +0000 (13:52 +0900)
commit1ff37de310e68814b282d9cbb3d1ef14512f9073
tree6360750fda200fb1f3a74c6cc08956b9fb3686aetree | snapshot
parent59a4c4ac98fb6d83613fff35f84c7057c21efc8ccommit | diff
[rnncell] implement multiout for rnncell

 - Calculate derivative inlcude duplicated process with calculate gradient.
   Needs to be optimized.
 - Remove hidden state and get previous hidden state from input
 - Remove timestep property
 - Disabled rnncell unittest. Proper unittest will be followed.

Signed-off-by: hyeonseok lee <hs89.lee@samsung.com>
nntrainer/compiler/recurrent_realizer.cpp diff | blob | history
nntrainer/layers/grucell.cpp diff | blob | history
nntrainer/layers/lstmcell.cpp diff | blob | history
nntrainer/layers/rnncell.cpp diff | blob | history
nntrainer/layers/rnncell.h diff | blob | history
test/unittest/layers/unittest_layers_rnncell.cpp diff | blob | history
test/unittest/models/unittest_models_recurrent.cpp diff | blob | history
Domain: Machine Learning / ML Framework;