search

HimaxWiseEyePlus / himax_tflm

Apache License 2.0
24 stars 15 forks source link

LSTM Problem. #12

Closed cjh8817676 closed 2 years ago

cjh8817676 commented 2 years ago

Is it possible to deploy LSTM layer on Himax We-i Plus Board with embarc_mli at https://github.com/foss-for-synopsys-dwc-arc-processors/embarc_mli.

I found that after downloading related third party, and I found that there is a compiled file of embarc_mli in path : ./third_party/arc_mli_package/bin/himax_arcem9d_r16/release called libmli.a .

kris-himax commented 2 years ago

@cjh8817676
The tflite-micro has not supported LSTM op yet. You can reference here "https://github.com/tensorflow/tflite-micro/issues/995" . When you create your own model using tf.Keras, you can use LSTM layer by setting unrolling=true. (Reference to "https://www.tensorflow.org/api_docs/python/tf/keras/layers/LSTM")