Why does LSTM layer do not support peephole?
If I have a LSTM model with peephole, can I use cudnn when I do inference?
Related topics
Topic | Replies | Views | Activity | |
---|---|---|---|---|
cuDNN6.0: LSTM with peephole connection | 5 | 2116 | September 21, 2017 | |
Use TenserRT2.1 for LSTM layer with peephole and projection | 0 | 480 | September 21, 2017 | |
layer norm for cudnn lstm | 0 | 953 | June 28, 2018 | |
ImportError: cannot import name 'CuDNNLSTM' from 'tensorflow.keras.layers' | 3 | 15451 | September 25, 2020 | |
API Reference :: NVIDIA Deep Learning cuDNN Documentation | 1 | 724 | January 4, 2021 | |
CuDNN — Status Not Intitialized (Keras/TensorFlow + Nvidia P100 + Linux) | 1 | 1218 | October 23, 2018 | |
How to implement a complete LSTM network with cudnn? | 5 | 3560 | January 13, 2020 | |
3D convolution/deconvolution support in TensorRT 3.0 ? | 4 | 1679 | February 20, 2019 | |
Question about cuDNN kernels | 0 | 397 | June 7, 2019 | |
Supported cudnn layers for int8 inference | 1 | 1078 | October 18, 2017 |