improv evignettes

.gitignore

@@ -9,3 +9,4 @@ mlr3torch*.tgz
 *~
 docs
 inst/doc
+*.html

DESCRIPTION

@@ -26,7 +26,11 @@ Authors@R:
              role = "ctb",
              email = "[email protected]",
              comment = c(ORCID = "0000-0001-8867-762X")))
-Description: Provides torch models as learners for the mlr3 ecosystem.
+Description: Deep Learning library that extends the mlr3 framework by building
+  upon the 'torch' package. It allows to conveniently build, train,
+  and evaluate deep learning models without having to worry about low level
+  details. Custom architectures can be created using the graph language
+  defined in 'mlr3pipelines'.
 License: LGPL (>= 3)
 Depends:
     mlr3 (>= 0.16.0),

R/CallbackSetHistory.R

@@ -70,8 +70,10 @@ CallbackSetHistory = R6Class("CallbackSetHistory",
         stopf("No eligible measures to plot for set '%s'.", set)
       }
 
+
+      epoch = score = measure = NULL
       if (ncol(data) == 2L) {
-        ggplot2::ggplot(data = data, ggplot2::aes_string(x = "epoch", y = measures)) +
+        ggplot2::ggplot(data = data, ggplot2::aes(x = epoch, y = !!rlang::sym(measures))) +
           ggplot2::geom_line() +
           ggplot2::geom_point() +
           ggplot2::labs(
@@ -82,7 +84,7 @@ CallbackSetHistory = R6Class("CallbackSetHistory",
           theme
       } else {
         data = melt(data, id.vars = "epoch", variable.name = "measure", value.name = "score")
-        ggplot2::ggplot(data = data, ggplot2::aes_string(x = "epoch", y = "score", color = "measure")) +
+        ggplot2::ggplot(data = data, ggplot2::aes_string(x = epoch, y = score, color = measure)) +
           viridis::scale_color_viridis(discrete = TRUE) +
           ggplot2::geom_line() +
           ggplot2::geom_point() +

R/ModelDescriptor.R

@@ -43,6 +43,7 @@
 #' @family Model Configuration
 #' @family Graph Network
 #' @return (`ModelDescriptor`)
+#' @export
 ModelDescriptor = function(graph, ingress, task, optimizer = NULL, loss = NULL, callbacks = NULL, .pointer = NULL,
   .pointer_shape = NULL) {
   assert_r6(graph, "Graph")
@@ -95,7 +96,7 @@ print.ModelDescriptor = function(x, ...) {
   catn(sprintf("<ModelDescriptor: %d ops>", length(x$graph$pipeops)))
   catn(str_indent("* Ingress: ", ingress_shapes))
   catn(str_indent("* Task: ", paste0(x$task$id, " [", x$task$task_type, "]")))
-  catn(str_indent("* Callbacks: ", if (!is.null(x$callbacks)) as_short_string(map_chr(x$callbacks, "label"), 100L) else "N/A")) # nolint
+  catn(str_indent("* Callbacks: ", if (!is.null(x$callbacks) && length(x$callbacks)) as_short_string(map_chr(x$callbacks, "label"), 100L) else "N/A")) # nolint
   catn(str_indent("* Optimizer: ", if (!is.null(x$optimizer)) as_short_string(x$optimizer$label) else "N/A"))
   catn(str_indent("* Loss: ", if (!is.null(x$loss)) as_short_string(x$loss$label) else "N/A"))
   catn(str_indent("* .pointer: ", if (is.null(x$.pointer)) "" else { # nolint

README.Rmd

@@ -20,7 +20,11 @@ lgr::get_logger("mlr3")$set_threshold("warn")
 
 
 
-# mlr3torch
+# mlr3torch <img src="man/figures/logo.png" align="right" width = "120" />
+
+Package website: [dev](https://mlr3torch.mlr-org.com/)
+
+Deep Learning with torch and mlr3.
 
 <!-- badges: start -->
 [![Lifecycle: experimental](https://img.shields.io/badge/lifecycle-experimental-orange.svg)](https://lifecycle.r-lib.org/articles/stages.html#experimental)
@@ -30,15 +34,12 @@ lgr::get_logger("mlr3")$set_threshold("warn")
 [![Mattermost](https://img.shields.io/badge/chat-mattermost-orange.svg)](https://lmmisld-lmu-stats-slds.srv.mwn.de/mlr_invite/)
 <!-- badges: end -->
 
-Deep Learning with torch and mlr3.
-
 ## Installation
 
 ```{r eval = FALSE}
 remotes::install_github("mlr-org/mlr3torch")
 ```
 
-
 ## Status
 
 `mlr3torch` is currently still unstable and many things are missing.
@@ -49,7 +50,7 @@ Not everything will work yet and the API might change without notice.
 
 `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 connection `PipeOpTorch` operators in an `mlr3pipelines::Graph`.
+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`.
 

README.md

@@ -1,7 +1,11 @@
 
 <!-- README.md is generated from README.Rmd. Please edit that file -->
 
-# mlr3torch
+# mlr3torch <img src="man/figures/logo.png" align="right" width = "120" />
+
+Package website: [dev](https://mlr3torch.mlr-org.com/)
+
+Deep Learning with torch and mlr3.
 
 <!-- badges: start -->
 
@@ -14,8 +18,6 @@ status](https://www.r-pkg.org/badges/version/mlr3torch)](https://CRAN.R-project.
 [![Mattermost](https://img.shields.io/badge/chat-mattermost-orange.svg)](https://lmmisld-lmu-stats-slds.srv.mwn.de/mlr_invite/)
 <!-- badges: end -->
 
-Deep Learning with torch and mlr3.
-
 ## Installation
 
 ``` r
@@ -34,7 +36,7 @@ everything will work yet and the API might change without notice.
 [`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 connection
+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)

_pkgdown.yml → pkgdown/_pkgdown.yml

@@ -25,13 +25,19 @@ navbar:
     mattermost:
       icon: fa fa-comments
       href: https://lmmisld-lmu-stats-slds.srv.mwn.de/mlr_invite/
+    github:
+      icon: fa fa-github
+      href: https://github.com/mlr-org/mlr3extralearners
     book:
       text: mlr3book
       icon: fa fa-link
       href: https://mlr3book.mlr-org.com
     stackoverflow:
       icon: fab fa-stack-overflow
       href: https://stackoverflow.com/questions/tagged/mlr3
+    rss:
+      icon: fa-rss
+      href: https://mlr-org.com/
 
 articles:
 - title: Get Started
@@ -41,7 +47,7 @@ articles:
 - title: Image Classification
   navbar: ~
   contents:
-  - image_classification
+  - articles/image_classification
 - title: Building a Neural Network using PipeOps
   navbar: ~
   contents:

tests/testthat/test_TorchDescriptor.R

@@ -11,7 +11,7 @@ test_that("TorchDescriptor basic checks", {
   expect_identical(descriptor$generator, nn_mse_loss)
   expect_identical(descriptor$id, "mse")
   expect_identical(descriptor$param_set$ids(), "reduction")
-  expect_set_equal(descriptor$packages, c("R6", "mlr3torch", "torch"))
+  expect_set_equal(descriptor$packages, c("R6", "torch"))
   expect_identical(descriptor$man, "torch::nn_mse_loss")
 
   expect_class(descriptor, "TorchDescriptor")
@@ -22,7 +22,7 @@ test_that("TorchDescriptor basic checks", {
     "<TorchDescriptor:mse> MSE Loss",
     "* Generator: nn_mse_loss",
     "* Parameters: list()",
-    "* Packages: R6,torch,mlr3torch"
+    "* Packages: R6,torch"
   )
   expect_identical(observed, expected)
 

vignettes/articles/image_classification.Rmd

@@ -15,12 +15,12 @@ In `mlr3torch`, image data is represented using the `mlr3torch::imageuri` class.
 It is essentially a character vector, containing paths to images on the file system.
 When listing the available task feature types (after loading the `mlr3torch` package), we can see that this class is available.
 
-```{r}
+```{r, message = FALSE}
 library(mlr3torch)
 mlr_reflections$task_feature_types
 ```
 
-Creating a vector (in this case of length 1) is as simple as passing the image paths to the `imageuri` function.
+Creating a vector (in this case of length 1) is as simple as passing the image paths to the `imageuri()` function.
 
 ```{r}
 image_vec = imageuri("/path/to/your/image")
@@ -30,20 +30,17 @@ For the processing of images, `mlr3torch` relies mostly on the functionality pro
 As an example task, we will use the "tiny imagenet" dataset, which is a subset of the [ImageNet](http://www.image-net.org/) dataset.
 It consists of 200 classes with 500 training images each.
 The goal is to predict the class of an image from the pixels.
-For more information you can access the tasks's help page.
+For more information you can access the tasks's help page by calling `$help()`.
 
 ```{r setup}
-set.seed(314)
-library(mlr3torch)
-
 tsk_tiny = tsk("tiny_imagenet")
 tsk_tiny
 ```
 
 The first time this task is accessed, the data is downloaded from the internet.
 In order to download the dataset only once, you can set the `mlr3torch.cache` option to either `TRUE` or a specific path to be used as the cache folder.
 
-```{r, eval = FALSE}
+```{r}
 options(mlr3torch.cache = TRUE)
 ```
 
@@ -61,5 +58,6 @@ alexnet = lrn("classif.alexnet")
 alexnet
 ```
 
-We can now train this learner like any other learner on the task at hand, while `mlr3torch` internally creates a dataloader from the image paths.
+We could now train this learner like any other learner on the task at hand, while `mlr3torch` internally creates a dataloader from the image paths.
+We could also download and use predefined weights, by specifying the `pretrained` parameter to `TRUE`.
 For computational reasons, we cannot demonstrate the actual training of the learner in this article.

vignettes/get_started.Rmd

@@ -21,7 +21,7 @@ We assume that you are familiar with the `mlr3` framework, for a detailed descri
 As a first example, we will train a simple multi-layer perceptron (MLP) on the well-known "mtcars" task.
 We first set a seed for reproducibility, load the library and construct the task.
 
-```{r}
+```{r, message = FALSE}
 set.seed(314)
 library(mlr3torch)
 task = tsk("mtcars")