multi birnn layer support???how to split input??

[ghost]

tensorlayer support deepbirnn layer????

how to use it??

multi birnn layer support???how to split input??

[zsdonghao]

@guoying1030 both BiRNNLayer and BiDynamicRNNLayer have a args n_layer, you can just simply set the number of RNN layers .

Alternatively, you can stack layers like this example shows: https://github.com/zsdonghao/tensorlayer/blob/master/example/tutorial_ptb_lstm_state_is_tuple.py

[ghost]

@zsdonghao ,, I would like to use multilayer birnn, and should meet the following conditions:

1. Each layer uses batch normal, and clip relu to 20 2.to expect the input coming from the same source of separate Sources？

What should I do? Use tensorlayer

[ghost]

@zsdonghao
I would like to use tensorlayer realize neon inside the function of Birnn, neon inside the BiRnn you can set split_input, the meaning of this parameter is: split_inputs (bool): to expect the input coming from the same source of separate Sources, but tensorlayer which birnn inside, I do not know how to set up this parameter? ?? This parameter from the code inside the neon see, assuming that the input is [32,1000,2304], respectively, on behalf of the [batch_size, n_step, hidden_size], then how split_input parameter is set to true, the function is running, [32,1000,1152] is used for forward propagation operations, and [32,1000,1152] is used for backward propagation operations. Then tensorlayer inside how to set up?

[wagamamaz]

close due to repeated issue