search

zhongkaifu / RNNSharp

RNNSharp is a toolkit of deep recurrent neural network which is widely used for many different kinds of tasks, such as sequence labeling, sequence-to-sequence and so on. It's written by C# language and based on .NET framework 4.6 or above versions. RNNSharp supports many different types of networks, such as forward and bi-directional network, sequence-to-sequence network, and different types of layers, such as LSTM, Softmax, sampled Softmax and others.
BSD 3-Clause "New" or "Revised" License
285 stars 91 forks source link

Possible combinations of various RNNs types #17

Closed My-Khan closed 8 years ago

My-Khan commented 8 years ago

Hello, i am little bit confused about the various combinations of model types and possible direction e.g forward and bidirectional. i am new to RNN architecture so please guide urgently. is the following combinations are valid ? if not then please explain little bit. thanks in advance. With Single Hidden Layer e.g –layersize 200 Bi-Directional LSTM-RNN without CRF Bi-Directional LSTM-RNN with CRF Bi-Directional BPTT-RNN without CRF Bi-Directional BPTT-RNN with CRF Forward LSTM-RNN without CRF Forward LSTM-RNN with CRF Forward BPTT-RNN without CRF Forward BPTT-RNN with CRF With Two Hidden Layer (deep layers) e.g –layersize 200,100 Bi-Directional LSTM-RNN without CRF Bi-Directional LSTM-RNN with CRF Bi-Directional BPTT-RNN without CRF Bi-Directional BPTT-RNN with CRF Forward LSTM-RNN without CRF Forward LSTM-RNN with CRF Forward BPTT-RNN without CRF Forward BPTT-RNN with CRF

zhongkaifu commented 8 years ago

Yes, they are valid. You could set different parameter values to train them.

My-Khan commented 8 years ago

Got it, many Thanks