iiinova — where your digital self keeps evolving

sebffischer.bsky.social @sebffischer.bsky.social · 06/10/2026

The cool thing about it is that you can just write the core algorithm in R, and anvl takes care of compiling it for either NVIDIA GPU or CPU! This also makes it super extensible, as you can easily go beyond hard-coded options in some predefined C++ library!

010

sebffischer.bsky.social @sebffischer.bsky.social · 06/10/2026

One goal we (with @louisaslett.bsky.social) have with anvl (github.com/r-xla/anvl) is to develop a new way to do Bayesian Inference in R, including easy-to use GPU acceleration. There are many things to do, so if you are looking for an open source project to contribute, this is your chance! #rstats

github.com

GitHub - r-xla/anvl: Accelerated array computing and code transformations for R

Accelerated array computing and code transformations for R - r-xla/anvl

193

Reposted by @sebffischer.bsky.social

Hugh @hughagraham.bsky.social · 05/10/2026

{anvl} is an #rstats game changer. Don't sleep on it!

072

Reposted by @sebffischer.bsky.social

sebffischer.bsky.social @sebffischer.bsky.social · 30/09/2026

Yesterday, we released a new version of anvl: github.com/r-xla/anvl. You can use it to implement numerical algorithms in R and run it on CPU and GPU. The R Code is compiled, so you do not have interpreter overhead, and it also allows you to automatically compute gradients #rstats

github.com

GitHub - r-xla/anvl: Accelerated array computing and code transformations for R

Accelerated array computing and code transformations for R - r-xla/anvl

0177

sebffischer.bsky.social @sebffischer.bsky.social · 30/09/2026

github.com

GitHub - r-xla/anvl: Accelerated array computing and code transformations for R

Accelerated array computing and code transformations for R - r-xla/anvl

0177

sebffischer.bsky.social @sebffischer.bsky.social · 18/08/2026

Note that {anvl} in principle also supports adding custom ops (C functions or CUDA functions), so almost anything can in principle be added to anvl. This mechanism is only partly implemented and not very ergonomic yet. If you are missing something that seems to require a custom op, let me know.

011

Reposted by @sebffischer.bsky.social

Hugh @hughagraham.bsky.social · 08/08/2026

Working on a new raster engine for #rstats, {garry}, powered by {gdalraster}, {anvl} and {mirai}. Still messy as hell but already the performance is pretty exciting. Only for the brave right now but ideas and opinions are still welcome! github.com/belian-earth... #rspatial #remotesensing

2259

sebffischer.bsky.social @sebffischer.bsky.social · 15/01/2025

A new package of the mlr3 family #mlr3inferr made it to CRAN. It allows to construct confidence intervals for the generalization error. You can check it out here: github.com/mlr-org/mlr3... #rstats.

github.com

GitHub - mlr-org/mlr3inferr: Statistical methods for inference on the generalization error

Statistical methods for inference on the generalization error - mlr-org/mlr3inferr

050