Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/mlr-org/mlr3torch
Deep learning framework for the mlr3 ecosystem based on torch
https://github.com/mlr-org/mlr3torch
data-science deep-learning machine-learning mlr3 r r-package torch
Last synced: 1 day ago
JSON representation
Deep learning framework for the mlr3 ecosystem based on torch
- Host: GitHub
- URL: https://github.com/mlr-org/mlr3torch
- Owner: mlr-org
- License: other
- Created: 2021-10-12T15:20:10.000Z (about 3 years ago)
- Default Branch: main
- Last Pushed: 2024-10-22T11:22:05.000Z (16 days ago)
- Last Synced: 2024-10-23T09:46:45.983Z (15 days ago)
- Topics: data-science, deep-learning, machine-learning, mlr3, r, r-package, torch
- Language: R
- Homepage: https://mlr3torch.mlr-org.com
- Size: 49.3 MB
- Stars: 38
- Watchers: 7
- Forks: 7
- Open Issues: 43
-
Metadata Files:
- Readme: README.Rmd
- Changelog: NEWS.md
- License: LICENSE.md
Awesome Lists containing this project
README
---
output: github_document
---
```{r, include = FALSE}
knitr::opts_chunk$set(
collapse = TRUE,
cache = FALSE,
comment = "#>",
fig.path = "man/figures/README-",
out.width = "100%"
)
set.seed(123)
library("mlr3torch")
lgr::get_logger("mlr3")$set_threshold("warn")
```
# mlr3torch 
Package website: [release](https://mlr3torch.mlr-org.com/) | [dev](https://mlr3torch.mlr-org.com/)
Deep Learning with torch and mlr3.
[](https://lifecycle.r-lib.org/articles/stages.html#experimental)
[](https://github.com/mlr-org/mlr3torch/actions/workflows/r-cmd-check.yml)
[](https://CRAN.R-project.org/package=mlr3torch)
[](https://stackoverflow.com/questions/tagged/mlr3)
[](https://lmmisld-lmu-stats-slds.srv.mwn.de/mlr_invite/)
## Installation
```{r eval = FALSE}
# Install from CRAN
install.packages("mlr3torch")
# Install the development version from GitHub:
pak::pak("mlr-org/mlr3torch")
```
Afterwards, you also need to run the command below:
```{r, eval = FALSE}
torch::install_torch()
```
More information about installing `torch` can be found [here](https://torch.mlverse.org/docs/articles/installation.html).
## What is mlr3torch?
`mlr3torch` is a deep learning framework for the [`mlr3`](https://mlr-org.com) ecosystem built on top of [`torch`](https://torch.mlverse.org/).
It allows to easily build, train and evaluate deep learning models in a few lines of codes, without needing to worry about low-level details.
Off-the-shelf learners are readily available, but custom architectures can be defined by connecting `PipeOpTorch` operators in an `mlr3pipelines::Graph`.
Using predefined learners such as a simple multi layer perceptron (MLP) works just like any other mlr3 `Learner`.
```{r}
library(mlr3torch)
learner_mlp = lrn("classif.mlp",
# defining network parameters
activation = nn_relu,
neurons = c(20, 20),
# training parameters
batch_size = 16,
epochs = 50,
device = "cpu",
# Proportion of data to use for validation
validate = 0.3,
# Defining the optimizer, loss, and callbacks
optimizer = t_opt("adam", lr = 0.1),
loss = t_loss("cross_entropy"),
callbacks = t_clbk("history"), # this saves the history in the learner
# Measures to track
measures_valid = msrs(c("classif.logloss", "classif.ce")),
measures_train = msrs(c("classif.acc")),
# predict type (required by logloss)
predict_type = "prob"
)
```
Below, we train this learner on the sonar example task:
```{r}
learner_mlp$train(tsk("sonar"))
```
Next, we construct the same architecture using `PipeOpTorch` objects.
The first pipeop -- a `PipeOpTorchIngress` -- defines the entrypoint of the network.
All subsequent pipeops define the neural network layers.
```{r}
architecture = po("torch_ingress_num") %>>%
po("nn_linear", out_features = 20) %>>%
po("nn_relu") %>>%
po("nn_head")
```
To turn this into a learner, we configure the loss, optimizer, callbacks as well as the training arguments.
```{r}
graph_mlp = architecture %>>%
po("torch_loss", loss = t_loss("cross_entropy")) %>>%
po("torch_optimizer", optimizer = t_opt("adam", lr = 0.1)) %>>%
po("torch_callbacks", callbacks = t_clbk("history")) %>>%
po("torch_model_classif",
batch_size = 16, epochs = 50, device = "cpu")
graph_lrn = as_learner(graph_mlp)
```
To work with generic tensors, the `lazy_tensor` type can be used.
It wraps a `torch::dataset`, but allows to preprocess the data (lazily) using `PipeOp` objects.
Below, we flatten the MNIST task, so we can then train a multi-layer perceptron on it.
Note that this does *not* transform the data in-memory, but is only applied when the data is actually loaded.
```{r}
# load the predefined mnist task
mnist = tsk("mnist")
mnist$head(3L)
# Flatten the images
flattener = po("trafo_reshape", shape = c(-1, 28 * 28))
mnist_flat = flattener$train(list(mnist))[[1L]]
mnist_flat$head(3L)
```
To actually access the tensors, we can call `materialize()`.
We only show a slice of the resulting tensor for readability:
```{r}
materialize(
mnist_flat$data(1:2, cols = "image")[[1L]],
rbind = TRUE
)[1:2, 1:4]
```
Below, we define a more complex architecture that has one single input which is a `lazy_tensor`.
For that, we first define a single residual block:
```{r}
layer = list(
po("nop"),
po("nn_linear", out_features = 50L) %>>%
po("nn_dropout") %>>% po("nn_relu")
) %>>% po("nn_merge_sum")
```
Next, we create a neural network that takes as input a `lazy_tensor` (`po("torch_ingress_ltnsr")`).
It first applies a linear layer and then repeats the above layer using the special `PipeOpTorchBlock`, followed by the network's head.
After that, we configure the loss, optimizer and the training parameters.
Note that `po("nn_linear_0")` is equivalent to `po("nn_linear", id = "nn_linear_0")` and we need this here to avoid ID clashes with the linear layer from `po("nn_block")`.
```{r}
deep_network = po("torch_ingress_ltnsr") %>>%
po("nn_linear_0", out_features = 50L) %>>%
po("nn_block", layer, n_blocks = 5L) %>>%
po("nn_head") %>>%
po("torch_loss", loss = t_loss("cross_entropy")) %>>%
po("torch_optimizer", optimizer = t_opt("adam")) %>>%
po("torch_model_classif",
epochs = 100L, batch_size = 32
)
```
Next, we prepend the preprocessing step that flattens the images so we can directly apply this learner to the unflattened MNIST task.
```{r}
deep_learner = as_learner(
flattener %>>% deep_network
)
deep_learner$id = "deep_network"
```
In order to keep track of the performance during training, we use 20% of the data and evaluate it using classification accuracy.
```{r}
set_validate(deep_learner, 0.2)
deep_learner$param_set$set_values(
torch_model_classif.measures_valid = msr("classif.acc")
)
```
```{r, include = FALSE}
# so it renders faster
deep_learner$param_set$values$torch_model_classif.epochs = 1L
mnist$filter(1:5)
```
All that is left is to train the learner:
```{r}
deep_learner$train(mnist)
```
## Feature Overview
* Off-the-shelf architectures are readily available as `mlr3::Learner`s.
* Currently, supervised regression and classification is supported.
* Custom learners can be defined using the `Graph` language from `mlr3pipelines`.
* The package supports tabular data, as well as generic tensors via the `lazy_tensor` type.
* Multi-modal data can be handled conveniently, as `lazy_tensor` objects can be stored alongside tabular data.
* It is possible to customize the training process via (predefined or custom) callbacks.
* The package is fully integrated into the `mlr3` ecosystem.
* Neural network architectures, as well as their hyperparameters can be easily tuned via `mlr3tuning` and friends.
## Documentation
* Start by reading one of the vignettes on the package website!
## Contributing:
* To run the tests one needs to set the environment variable `TEST_TORCH = 1`, e.g. by adding it to `.Renviron`.
## Acknowledgements
* Without the great R package `torch` none of this would have been possible.
* The names for the callback stages are taken from [luz](https://mlverse.github.io/luz/), another high-level deep learning framework for R `torch`.
* Building neural networks using `PipeOpTorch` operators is inspired by [keras](https://keras.io/).
## Bugs, Questions, Feedback
*mlr3torch* is a free and open source software project that
encourages participation and feedback. If you have any issues,
questions, suggestions or feedback, please do not hesitate to open an
“issue” about it on the GitHub page\!
In case of problems / bugs, it is often helpful if you provide a
“minimum working example” that showcases the behaviour (but don’t
worry about this if the bug is obvious).
Please understand that the resources of the project are limited:
response may sometimes be delayed by a few days, and some feature
suggestions may be rejected if they are deemed too tangential to the
vision behind the project.