torchvision icon indicating copy to clipboard operation
torchvision copied to clipboard

R interface to torchvision

Results 26 torchvision issues
Sort by recently updated
recently updated
newest added

I just stumbled upon an oddity with MNIST. The labels seem to be off by one: Annotated labels are 1..10 rather than 0..9 and indeed visualizing the images shows that...

The standard dataset_generator for MNIST dataset does not include a `.getbatch()` method and, as a result, getting a batch is quite slow, at least on CPU. ``` # dataset root...

Any launching of model_resnet functions throws error "Error in !self$..refer_to_state_dict.. : invalid argument type" Seems like some piece of reference information is missed in models.

Error when Loading the Package: library(torchvisionlib) using R Console Can see torchvisionlib.dll in the folder; but just can't load it. > library(torchvisionlib) **trying URL 'https://github.com/mlverse/torchvisionlib/releases/download/v0.4.0.9000/torchvisionlib-0.4.0.9000+cpu-win64.zip' Content type 'application/octet-stream' length 2834668...

Hi! I see that R torch support mnist and cifar data sets, but I was wondering if there are plans to support the other data sets which are present in...

``` r torchvision::transform_random_rotation(torch::torch_randn(3, 10, 10), degrees = 3) #> Error in if (mode == "bilinear") {: missing value where TRUE/FALSE needed ``` Created on 2023-12-01 with [reprex v2.0.2](https://reprex.tidyverse.org)

Maybe I don't fully understand the `transform_rotate()` function, but below it yields a (for me) unexpected output shape. ``` r library(torchvision) library(torch) #> Warning: package 'torch' was built under R...

Below, the file `tiny-imagenet-200.zip` should be deleted I believe after it is unzipped. ``` r dir = tempfile() torchvision::tiny_imagenet_dataset(dir, download = TRUE) #> Downloding tiny imagenet dataset! #> Download complete....

Hello, thanks again for the great work on making torch available to R users! I was wondering if you are interested in allowing users to specify the number of input...

Hey there, I was wondering if I'm doing anything wrong, or if the transform_grayscale functionality is actually not implemented for any class? Backtrace: ▆ 1. ├─torchvision::transform_grayscale(img = img, num_output_channels =...