Ecosyste.ms: Awesome

An open API service indexing awesome lists of open source software.

Awesome Lists | Featured Topics | Projects

https://github.com/szymonmaszke/torchlayers

Shape and dimension inference (Keras-like) for PyTorch layers and neural networks
https://github.com/szymonmaszke/torchlayers

deep-learning framework inference keras layers library machine-learning pytorch shape sota

Last synced: about 1 month ago
JSON representation

Shape and dimension inference (Keras-like) for PyTorch layers and neural networks

Awesome Lists containing this project

README

        

![torchlayers Logo](https://github.com/szymonmaszke/torchlayers/blob/master/assets/banner.png)

--------------------------------------------------------------------------------

| Version | Docs | Tests | Coverage | Style | PyPI | Python | PyTorch | Docker |
|---------|------|-------|----------|-------|------|--------|---------|--------|
| [![Version](https://img.shields.io/static/v1?label=&message=0.1.1&color=377EF0&style=for-the-badge)](https://github.com/szymonmaszke/torchlayers/releases) | [![Documentation](https://img.shields.io/static/v1?label=&message=docs&color=EE4C2C&style=for-the-badge)](https://szymonmaszke.github.io/torchlayers/) | ![Tests](https://img.shields.io/github/workflow/status/szymonmaszke/torchlayers/test?label=%20&style=for-the-badge) | [![codecov](https://codecov.io/gh/szymonmaszke/torchlayers/branch/master/graph/badge.svg?token=GbZmdqbTWM)](https://codecov.io/gh/szymonmaszke/torchlayers) | [![codebeat badge](https://codebeat.co/badges/0e3d33b0-95a4-429c-8692-881a4ffeac6b)](https://codebeat.co/projects/github-com-szymonmaszke-torchlayers-master) | [![PyPI](https://img.shields.io/static/v1?label=&message=PyPI&color=377EF0&style=for-the-badge)](https://pypi.org/project/torchlayers/) | [![Python](https://img.shields.io/static/v1?label=&message=>=3.7&color=377EF0&style=for-the-badge&logo=python&logoColor=F8C63D)](https://www.python.org/) | [![PyTorch](https://img.shields.io/static/v1?label=&message=>=1.3.0&color=EE4C2C&style=for-the-badge)](https://pytorch.org/) | [![Docker](https://img.shields.io/static/v1?label=&message=docker&color=309cef&style=for-the-badge)](https://hub.docker.com/r/szymonmaszke/torchlayers) |

[__torchlayers__](https://szymonmaszke.github.io/torchlayers/) is a library based on [__PyTorch__](https://pytorch.org/)
providing __automatic shape and dimensionality inference of `torch.nn` layers__ + additional
building blocks featured in current SOTA architectures (e.g. [Efficient-Net](https://arxiv.org/abs/1905.11946)).

Above requires no user intervention (except single call to `torchlayers.build`)
similarly to the one seen in [__Keras__](https://www.tensorflow.org/guide/keras).

### Main functionalities:

* __Shape inference__ for most of `torch.nn` module (__convolutional, recurrent, transformer, attention and linear layers__)
* __Dimensionality inference__ (e.g. `torchlayers.Conv` working as `torch.nn.Conv1d/2d/3d` based on `input shape`)
* __Shape inference of custom modules__ (see examples section)
* __Additional [Keras-like](https://www.tensorflow.org/guide/keras) layers__ (e.g. `torchlayers.Reshape` or `torchlayers.StandardNormalNoise`)
* __Additional SOTA layers__ mostly from ImageNet competitions
(e.g. [PolyNet](https://arxiv.org/abs/1608.06993),
[Squeeze-And-Excitation](https://arxiv.org/abs/1709.01507),
[StochasticDepth](www.arxiv.org/abs/1512.03385>))
* __Useful defaults__ (`"same"` padding and default `kernel_size=3` for `Conv`, dropout rates etc.)
* __Zero overhead and [torchscript](https://pytorch.org/docs/stable/jit.html) support__

__Keep in mind this library works almost exactly like PyTorch originally__.
What that means is you can use `Sequential`, __define your own networks of any complexity using
`torch.nn.Module`__, create new layers with shape inference etc.

_See below to get some intuition about library_.

# Examples

For full functionality please check [__torchlayers documentation__](https://szymonmaszke.github.io/torchlayers/).
Below examples should introduce all necessary concepts you should know.

## Basic classifier

__All__ `torch.nn` modules can be used through `torchlayers` and __each module with input shape__
will be appropriately modified with it's input inferable counterpart.

```python
import torchlayers as tl

class Classifier(tl.Module):
def __init__(self):
super().__init__()
self.conv1 = tl.Conv2d(64, kernel_size=6)
self.conv2 = tl.Conv2d(128, kernel_size=3)
self.conv3 = tl.Conv2d(256, kernel_size=3, padding=1)
# New layer, more on that in the next example
self.pooling = tl.GlobalMaxPool()
self.dense = tl.Linear(10)

def forward(self, x):
x = torch.relu(self.conv1(x))
x = torch.relu(self.conv2(x))
x = torch.relu(self.conv3(x))
return self.dense(self.pooling(x))

# Pass model and any example inputs afterwards
clf = tl.build(Classifier(), torch.randn(1, 3, 32, 32))
```

Above `torchlayers.Linear(out_features=10)` is used. It is "equivalent" to
original PyTorch's `torch.nn.Linear(in_features=?, out_features=10)` where `in_features`
will be inferred from example input input during `torchlayers.build` call.

Same thing happens with `torch.nn.Conv2d(in_channels, out_channels, kernel_size, ...)`
which can be replaced directly by `tl.Conv2d(out_channels, kernel_size, ...)`.

__Just remember to pass example input through the network!__

## Simple image and text classifier in one!

* We will use single "model" for both tasks.
Firstly let's define it using `torch.nn` and `torchlayers`:

```python
import torch
import torchlayers as tl

# torch.nn and torchlayers can be mixed easily
model = torch.nn.Sequential(
tl.Conv(64), # specify ONLY out_channels
torch.nn.ReLU(), # use torch.nn wherever you wish
tl.BatchNorm(), # BatchNormNd inferred from input
tl.Conv(128), # Default kernel_size equal to 3
tl.ReLU(),
tl.Conv(256, kernel_size=11), # "same" padding as default
tl.GlobalMaxPool(), # Known from Keras
tl.Linear(10), # Output for 10 classes
)

print(model)
```

Above would give you model's summary like this (__notice question marks for not yet inferred values__):

```python
Sequential(
(0): Conv(in_channels=?, out_channels=64, kernel_size=3, stride=1, padding=same, dilation=1, groups=1, bias=True, padding_mode=zeros)
(1): ReLU()
(2): BatchNorm(num_features=?, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(3): Conv(in_channels=?, out_channels=128, kernel_size=3, stride=1, padding=same, dilation=1, groups=1, bias=True, padding_mode=zeros)
(4): ReLU()
(5): Conv(in_channels=?, out_channels=256, kernel_size=11, stride=1, padding=same, dilation=1, groups=1, bias=True, padding_mode=zeros)
(6): GlobalMaxPool()
(7): Linear(in_features=?, out_features=10, bias=True)
)
```

* Now you can __build__/instantiate your model with example input (in this case MNIST-like):

```python
mnist_model = tl.build(model, torch.randn(1, 3, 28, 28))
```

* Or if it's text classification you are after, same model could be built with different
`input shape` (e.g. for text classification using `300` dimensional pretrained embedding):

```python
# [batch, embedding, timesteps], first dimension > 1 for BatchNorm1d to work
text_model = tl.build(model, torch.randn(2, 300, 1))
```

* Finally, you can `print` both models after instantiation, provided below side
by-side for readability (__notice different dimenstionality, e.g. `Conv2d` vs `Conv1d` after `torchlayers.build`__):

```python
# TEXT CLASSIFIER MNIST CLASSIFIER

Sequential( Sequential(
(0): Conv1d(300, 64) (0): Conv2d(3, 64)
(1): ReLU() (1): ReLU()
(2): BatchNorm1d(64) (2): BatchNorm2d(64)
(3): Conv1d(64, 128) (3): Conv2d(64, 128)
(4): ReLU() (4): ReLU()
(5): Conv1d(128, 256) (5): Conv2d(128, 256)
(6): GlobalMaxPool() (6): GlobalMaxPool()
(7): Linear(256, 10) (7): Linear(256, 10)
) )
```

As you can see both modules "compiled" into original `pytorch` layers.

## Custom modules with shape inference capabilities

User can define any module and make it shape inferable with `torchlayers.infer`
function:

```python
# Class defined with in_features
# It might be a good practice to use _ prefix and Impl as postfix
# to differentiate from shape inferable version
class _MyLinearImpl(torch.nn.Module):
def __init__(self, in_features: int, out_features: int):
super().__init__()
self.weight = torch.nn.Parameter(torch.randn(out_features, in_features))
self.bias = torch.nn.Parameter(torch.randn(out_features))

def forward(self, inputs):
return torch.nn.functional.linear(inputs, self.weight, self.bias)

MyLinear = tl.infer(_MyLinearImpl)

# Build and use just like any other layer in this library
layer =tl.build(MyLinear(out_features=32), torch.randn(1, 64))
layer(torch.randn(1, 64))
```

By default `inputs.shape[1]` will be used as `in_features` value
during initial `forward` pass. If you wish to use different `index` (e.g. to infer using
`inputs.shape[3]`) use `MyLayer = tl.infer(_MyLayerImpl, index=3)` as a decorator.

## Autoencoder with inverted residual bottleneck and pixel shuffle

Please check code comments and [__documentation__](https://szymonmaszke.github.io/torchlayers/)
if needed. If you are unsure what autoencoder is you could see
[__this example blog post__](https://towardsdatascience.com/auto-encoder-what-is-it-and-what-is-it-used-for-part-1-3e5c6f017726).

Below is a convolutional denoising autoencoder example for `ImageNet`-like images.
Think of it like a demonstration of capabilities of different layers
and building blocks provided by `torchlayers`.

```python
# Input - 3 x 256 x 256 for ImageNet reconstruction
class AutoEncoder(torch.nn.Module):
def __init__(self):
super().__init__()
self.encoder = tl.Sequential(
tl.StandardNormalNoise(), # Apply noise to input images
tl.Conv(64, kernel_size=7),
tl.activations.Swish(), # Direct access to module .activations
tl.InvertedResidualBottleneck(squeeze_excitation=False),
tl.AvgPool(), # shape 64 x 128 x 128, kernel_size=2 by default
tl.HardSwish(), # Access simply through tl
tl.SeparableConv(128), # Up number of channels to 128
tl.InvertedResidualBottleneck(), # Default with squeeze excitation
torch.nn.ReLU(),
tl.AvgPool(), # shape 128 x 64 x 64, kernel_size=2 by default
tl.DepthwiseConv(256), # DepthwiseConv easier to use
# Pass input thrice through the same weights like in PolyNet
tl.Poly(tl.InvertedResidualBottleneck(), order=3),
tl.ReLU(), # all torch.nn can be accessed via tl
tl.MaxPool(), # shape 256 x 32 x 32
tl.Fire(out_channels=512), # shape 512 x 32 x 32
tl.SqueezeExcitation(hidden=64),
tl.InvertedResidualBottleneck(),
tl.MaxPool(), # shape 512 x 16 x 16
tl.InvertedResidualBottleneck(squeeze_excitation=False),
# Randomly switch off the last two layers with 0.5 probability
tl.StochasticDepth(
torch.nn.Sequential(
tl.InvertedResidualBottleneck(squeeze_excitation=False),
tl.InvertedResidualBottleneck(squeeze_excitation=False),
),
p=0.5,
),
tl.AvgPool(), # shape 512 x 8 x 8
)

# This one is more "standard"
self.decoder = tl.Sequential(
tl.Poly(tl.InvertedResidualBottleneck(), order=2),
# Has ICNR initialization by default after calling `build`
tl.ConvPixelShuffle(out_channels=512, upscale_factor=2),
# Shape 512 x 16 x 16 after PixelShuffle
tl.Poly(tl.InvertedResidualBottleneck(), order=3),
tl.ConvPixelShuffle(out_channels=256, upscale_factor=2),
# Shape 256 x 32 x 32
tl.Poly(tl.InvertedResidualBottleneck(), order=3),
tl.ConvPixelShuffle(out_channels=128, upscale_factor=2),
# Shape 128 x 64 x 64
tl.Poly(tl.InvertedResidualBottleneck(), order=4),
tl.ConvPixelShuffle(out_channels=64, upscale_factor=2),
# Shape 64 x 128 x 128
tl.InvertedResidualBottleneck(),
tl.Conv(256),
tl.Dropout(), # Defaults to 0.5 and Dropout2d for images
tl.Swish(),
tl.InstanceNorm(),
tl.ConvPixelShuffle(out_channels=32, upscale_factor=2),
# Shape 32 x 256 x 256
tl.Conv(16),
tl.Swish(),
tl.Conv(3),
# Shape 3 x 256 x 256
)

def forward(self, inputs):
return self.decoder(self.encoder(inputs))
```

Now one can instantiate the module and use it with `torch.nn.MSELoss` as per usual.

```python
autoencoder = tl.build(AutoEncoder(), torch.randn(1, 3, 256, 256))
```

# Installation

## [pip]()

### Latest release:

```shell
pip install --user torchlayers
```

### Nightly:

```shell
pip install --user torchlayers-nightly
```

## [Docker](https://hub.docker.com/r/szymonmaszke/torchlayers)

__CPU standalone__ and various versions of __GPU enabled__ images are available
at [dockerhub](https://hub.docker.com/r/szymonmaszke/torchlayers/tags).

For CPU quickstart, issue:

```shell
docker pull szymonmaszke/torchlayers:18.04
```

Nightly builds are also available, just prefix tag with `nightly_`. If you are going for `GPU` image make sure you have
[nvidia/docker](https://github.com/NVIDIA/nvidia-docker) installed and it's runtime set.

# Contributing

If you find issue or would like to see some functionality (or implement one), please [open new Issue](https://help.github.com/en/articles/creating-an-issue) or [create Pull Request](https://help.github.com/en/articles/creating-a-pull-request-from-a-fork).