[GitHub] [incubator-mxnet] zixuanweeei commented on issue #15745: Memory layout in the LSTM operator
zixuanweeei commented on issue #15745: Memory layout in the LSTM operator URL: https://github.com/apache/incubator-mxnet/issues/15745#issuecomment-518461277 Feel free to directly mention me here if there is any question . BTW, we are working on integrating the LBR-GRU of MKL-DNN into MXNet. It will be completed in these days. This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] zixuanweeei commented on issue #15745: Memory layout in the LSTM operator
zixuanweeei commented on issue #15745: Memory layout in the LSTM operator URL: https://github.com/apache/incubator-mxnet/issues/15745#issuecomment-518454755 Is there any problem with the order? The native LSTM implementation of MXNet shares the same order of gates with that of MKL-DNN, but differs in the number of bias. And the gates order of their GRU implementations are different, which might be concerned. This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-mxnet] zixuanweeei commented on issue #15745: Memory layout in the LSTM operator
zixuanweeei commented on issue #15745: Memory layout in the LSTM operator URL: https://github.com/apache/incubator-mxnet/issues/15745#issuecomment-517971666 @eloi-loomai The memory layout of weights is: ``` L * H * ngates * H L * H * ngates * H L * nbias * H +--+--+-+ workptrweight_iter_nbias_n others weight_layer_n ``` So it should be `DType* bias_n = weight_iter_n + L * H * ngates * H;`. And it should be noticed that the [LSTM formula of MXNet](https://mxnet.incubator.apache.org/api/python/gluon/rnn.html#mxnet.gluon.rnn.LSTMCell) differs from [that of MKL-DNN](https://intel.github.io/mkl-dnn/dev_guide_rnn.html). MXNet has two parts of biases in each gate of RNN variants, while MKL-DNN only has a single bias, except for the bias of current memory content of LBR-GRU. This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services