readme.md

# [LabML Neural Networks](https://lab-ml.com/labml_nn/index.html)

![Screenshot](https://github.com/lab-ml/nn/blob/master/images/dqn.png)

This is a collection of simple PyTorch implementations of
neural networks and related algorithms.
These implementations are documented with explanations,
and the [website](https://lab-ml.com/labml_nn/index.html)
renders these as side-by-side formatted notes.
We believe these would help you understand these algorithms better.

We are actively maintaining this repo and adding new 
implementations.

## Modules

#### ✨ [Transformers](https://lab-ml.com/labml_nn/transformers)

[Transformers module](https://lab-ml.com/labml_nn/transformers)
contains implementations for
[multi-headed attention](https://lab-ml.com/labml_nn/transformers/mha.html)
and
[relative multi-headed attention](https://lab-ml.com/labml_nn/transformers/relative_mha.html).

* [kNN-LM: Generalization through Memorization](https://lab-ml.com/labml_nn/transformers/knn)
* [Feedback Transformer](https://lab-ml.com/labml_nn/transformers/feedback)

#### ✨ [Recurrent Highway Networks](https://lab-ml.com/labml_nn/recurrent_highway_networks)

#### ✨ [LSTM](https://lab-ml.com/labml_nn/lstm)

#### ✨ [HyperNetworks - HyperLSTM](https://lab-ml.com/labml_nn/hypernetworks/hyper_lstm.html)

#### ✨ [Capsule Networks](https://lab-ml.com/labml_nn/capsule_networks/)

#### ✨ [Generative Adversarial Networks](https://lab-ml.com/labml_nn/gan/)
* [GAN with a multi-layer perceptron](https://lab-ml.com/labml_nn/gan/simple_mnist_experiment.html)
* [GAN with deep convolutional network](https://lab-ml.com/labml_nn/gan/dcgan.html)
* [Cycle GAN](https://lab-ml.com/labml_nn/gan/cycle_gan.html)

#### ✨ [Sketch RNN](https://lab-ml.com/labml_nn/sketch_rnn/)

#### ✨ [Reinforcement Learning](https://lab-ml.com/labml_nn/rl/)
* [Proximal Policy Optimization](https://lab-ml.com/labml_nn/rl/ppo/) with
 [Generalized Advantage Estimation](https://lab-ml.com/labml_nn/rl/ppo/gae.html)
* [Deep Q Networks](https://lab-ml.com/labml_nn/rl/dqn/) with
 with [Dueling Network](https://lab-ml.com/labml_nn/rl/dqn/model.html),
 [Prioritized Replay](https://lab-ml.com/labml_nn/rl/dqn/replay_buffer.html)
 and Double Q Network.

#### ✨ [Optimizers](https://lab-ml.com/labml_nn/optimizers/)
* [Adam](https://lab-ml.com/labml_nn/optimizers/adam.html)
* [AMSGrad](https://lab-ml.com/labml_nn/optimizers/amsgrad.html)
* [Adam Optimizer with warmup](https://lab-ml.com/labml_nn/optimizers/adam_warmup.html)
* [Noam Optimizer](https://lab-ml.com/labml_nn/optimizers/noam.html)
* [Rectified Adam Optimizer](https://lab-ml.com/labml_nn/optimizers/radam.html)
* [AdaBelief Optimizer](https://lab-ml.com/labml_nn/optimizers/ada_belief.html)

### Installation

```bash
pip install labml_nn
```

### Citing LabML

If you use LabML for academic research, please cite the library using the following BibTeX entry.

```bibtex
@misc{labml,
 author = {Varuna Jayasiri, Nipun Wijerathne},
 title = {LabML: A library to organize machine learning experiments},
 year = {2020},
 url = {https://lab-ml.com/},
}
```