char-rnn

This code implements multi-layer Recurrent Neural Network (RNN, LSTM, and GRU) for training/sampling from character-level language models. The model learns to predict the probability of the next character in a sequence. In other words, the input is a single text file and the model learns to generate text like it.

The context of this code base is described in detail in my blog post. The project page that has a few pointers to some datasets.

This code was originally based on Oxford University Machine Learning class practical 6, which is in turn based on learning to execute code from Wojciech Zaremba. Chunks of it were also developed in collaboration with my labmate Justin Johnson.

Requirements

This code is written in Lua and requires Torch. Additionally, you need to install the nngraph and optim packages using LuaRocks which you will be able to do after installing Torch:

$ luarocks install nngraph 
$ luarocks install optim

If you'd like to use CUDA GPU computing, you'll first need to install the CUDA Toolkit, then the cutorch and cunn packages:

$ luarocks install cutorch
$ luarocks install cunn

If you'd like to use OpenCL GPU computing, you'll first need to install the cltorch and clnn packages, and then use the option -opencl 1 during training (cltorch issues):

$ luarocks install cltorch
$ luarocks install clnn

Usage

Data

All input data is stored inside the data/ directory. You'll notice that there is an example dataset included in the repo (in folder data/tinyshakespeare) which consists of a subset of works of Shakespeare. I'm providing a few more datasets on the project page.

Your own data: If you'd like to use your own data create a single file input.txt and place it into a folder in data/. For example, data/some_folder/input.txt. The first time you run the training script it will write two more convenience files into data/some_folder.

Note that if your data is too small (1MB is already considered very small) the RNN won't learn very effectively. Remember that it has to learn everything completely from scratch.

Conversely if your data is large (more than about 2MB), feel confident to increase rnn_size and train a bigger model (see details of training below). It will work significantly better. For example with 6MB you can easily go up to rnn_size 300 or even more. The biggest that fits on my GPU and that I've trained with this code is rnn_size 700 with num_layers 3 (2 is default).

Training

Start training the model using train.lua, for example:

$ th train.lua -data_dir data/some_folder -gpuid -1

The -data_dir flag is most important since it specifies the dataset to use. Notice that in this example we're also setting gpuid to -1 which tells the code to train using CPU, otherwise it defaults to GPU 0. There are many other flags for various options. Consult $ th train.lua -help for comprehensive settings. Here's another example:

$ th train.lua -data_dir data/some_folder -rnn_size 512 -num_layers 2 -dropout 0.5

While the model is training it will periodically write checkpoint files to the cv folder. The frequency with which these checkpoints are written is controlled with number of iterations, as specified with the eval_val_every option (e.g. if this is 1 then a checkpoint is written every iteration). The filename of these checkpoints contains a very imporatant number: the loss. For example, a checkpoint with filename lm_lstm_epoch0.95_2.0681.t7 indicates that at this point the model was on epoch 0.95 (i.e. it has almost done one full pass over the training data), and the loss on validation data was 2.0681. This number is very important because the lower it is, the better the checkpoint works. Once you start to generate data (discussed below), you will want to use the model checkpoint that has the lowest validation loss. Notice that this might not necessarily be the last checkpoint at the end of training (due to possible overfitting).

Another important quantities to be aware of are batch_size (call it B), seq_length (call it S), and the train_frac and val_frac settings. The batch size specifies how many streams of data are processed in parallel at one time. The sequence length specifies the length of each chunk, which is also the limit at which the gradients get clipped. For example, if seq_length is 20, then the gradient signal will never backpropagate more than 20 time steps, and the model might not find dependencies longer than this length in number of characters. At runtime your input text file has N characters, these first all get split into chunks of size BxS. These chunks then get allocated to three splits: train/val/test according to the frac settings. If your data is small, it's possible that with the default settings you'll only have very few chunks in total (for example 100). This is bad: In these cases you may want to decrease batch size or sequence length.

You can also init parameters from a previously saved checkpoint using init_from.

We can use these checkpoints to generate text (discussed next).

Sampling

Given a checkpoint file (such as those written to cv) we can generate new text. For example:

$ th sample.lua cv/some_checkpoint.t7 -gpuid -1

Make sure that if your checkpoint was trained with GPU it is also sampled from with GPU, or vice versa. Otherwise the code will (currently) complain. As with the train script, see $ th sample.lua -help for full options. One important one is (for example) -length 10000 which would generate 10,000 characters (default = 2000).

Temperature. An important parameter you may want to play with a lot is -temperature, which takes a number in range [0, 1] (notice 0 not included), default = 1. The temperature is dividing the predicted log probabilities before the Softmax, so lower temperature will cause the model to make more likely, but also more boring and conservative predictions. Higher temperatures cause the model to take more chances and increase diversity of results, but at a cost of more mistakes.

Priming. It's also possible to prime the model with some starting text using -primetext. This starts out the RNN with some hardcoded characters to warm it up with some context before it starts generating text.

Happy sampling!

Tips and Tricks

Monitoring Validation Loss vs. Training Loss

If you're somewhat new to Machine Learning or Neural Networks it can take a bit of expertise to get good models. The most important quantity to keep track of is the difference between your training loss (printed during training) and the validation loss (printed once in a while when the RNN is run on the validation data (by default every 1000 iterations)). In particular:

• If your training loss is much lower than validation loss then this means the network might be overfitting. Solutions to this are to decrease your network size, or to increase dropout. For example you could try dropout of 0.5 and so on.
• If your training/validation loss are about equal then your model is underfitting. Increase the size of your model (either number of layers or the raw number of neurons per layer)

Approximate number of parameters

The two most important parameters that control the model are rnn_size and num_layers. I would advise that you always use num_layers of either 2/3. The rnn_size can be adjusted based on how much data you have. The two important quantities to keep track of here are:

• The number of parameters in your model. This is printed when you start training.
• The size of your dataset. 1MB file is approximately 1 million characters.

These two should be about the same order of magnitude. It's a little tricky to tell. Here are some examples:

• I have a 100MB dataset and I'm using the default parameter settings (which currently print 150K parameters). My data size is significantly larger (100 mil >> 0.15 mil), so I expect to heavily underfit. I am thinking I can comfortably afford to make rnn_size larger.
• I have a 10MB dataset and running a 10 million parameter model. I'm slightly nervous and I'm carefully monitoring my validation loss. If it's larger than my training loss then I may want to increase dropout a bit.

Best models strategy

The winning strategy to obtaining very good models (if you have the compute time) is to always err on making the network larger (as large as you're willing to wait for it to compute) and then try different dropout values (between 0,1). Whatever model has the best validation performance (the loss, written in the checkpoint filename, low is good) is the one you should use in the end.

It is very common in deep learning to run many different models with many different hyperparameter settings, and in the end take whatever checkpoint gave the best validation performance.

By the way, the size of your training and validation splits are also parameters. Make sure you have a decent amount of data in your validation set or otherwise the validation performance will be noisy and not very informative.

License

MIT