Note that the master branch of Caffe supports LSTM now. (Jeff Donahue's implementation has been merged.)
This repo is no longer maintained.

Speed comparison (Titan X, 3-layer LSTM with 2048 units)

Jeff's code is more modularized, whereas this code is optimized for LSTM.
This code computes gradient w.r.t. recurrent weights with a single matrix computation.

• Batch size = 20, Length = 100

Code	Forward(ms)	Backward(ms)	Total (ms)
This code	248	291	539
Jeff's code	264	462	726

• Batch size = 4, Length = 100

Code	Forward(ms)	Backward(ms)	Total (ms)
This code	131	118	249
Jeff's code	140	290	430

• Batch size = 20, Length = 20

Code	Forward(ms)	Backward(ms)	Total (ms)
This code	49	59	108
Jeff's code	52	92	144

• Batch size = 4, Length = 20

Code	Forward(ms)	Backward(ms)	Total (ms)
This code	29	26	55
Jeff's code	30	61	91

Example

An example code is in /examples/lstm_sequence/.
In this code, LSTM network is trained to generate a predefined sequence without any inputs.
This experiment was introduced by Clockwork RNN.
Four different LSTM networks and shell scripts(.sh) for training are provided.
Each script generates a log file containing the predicted sequence and the true sequence.
You can use plot_result.m to visualize the result.
The result of four LSTM networks will be as follows:

• 1-layer LSTM with 15 hidden units for short sequence
• 1-layer LSTM with 50 hidden units for long sequence
• 3-layer deep LSTM with 7 hidden units for short sequence
• 3-layer deep LSTM with 23 hidden units for long sequence