Elektrine
EN
Log in Register
Paige Chat Timeline Communities Gallery Videos Email DNS VPN Uptime Kairo
Back to Timeline
Remote

sebffischer

@sebffischer@mastodon.social
mastodon 4.7.0-beta.1
  • Open on mastodon.social

PhD Student in Statistics / Machine Learning.
Working on mlr3, which is a machine learning framework in R.

Also working on a code transformation framework for R (jit compilation, autodiff): https://r-xla.github.io/anvl/

0 Followers
0 Following
24 Posts
Joined February 02, 2024

Posts

Open post
sebffischer
sebffischer @sebffischer@mastodon.social · May 07, 2026
sebffischer
@sebffischer@mastodon.social

PhD Student in Statistics / Machine Learning.
Working on mlr3, which is a machine learning framework in R. Also working on a code transformation framework for R (jit compilation, autodiff): https://r-xla.github.io/anvl/

mastodon.social

In an R package, I am linking against R's LAPACK but ideally would like to have access to the the single precision float routines, which are not bundled there (on windows) because R does not use them. Does anyone have a recommendation how to also get access to f32 routines on windows? #rstats

mastodon.social

Mastodon

1
2
4
0
Open post
sebffischer
sebffischer @sebffischer@mastodon.social · May 01, 2026
sebffischer
@sebffischer@mastodon.social

PhD Student in Statistics / Machine Learning.
Working on mlr3, which is a machine learning framework in R. Also working on a code transformation framework for R (jit compilation, autodiff): https://r-xla.github.io/anvl/

mastodon.social
Replying to @sebffischer@mastodon.social
One thing which might be nice is to have a ..(n) function that allows to skip n dimensions, so slicing the fourth dimension would become x[..(3), i, ..]
0
0
0
0
Open post
sebffischer
sebffischer @sebffischer@mastodon.social · May 01, 2026
sebffischer
@sebffischer@mastodon.social

PhD Student in Statistics / Machine Learning.
Working on mlr3, which is a machine learning framework in R. Also working on a code transformation framework for R (jit compilation, autodiff): https://r-xla.github.io/anvl/

mastodon.social
Replying to @sebffischer@mastodon.social
as more and more code is produced by LLMs I think it becomes even more important for code to be intuitive and readable.
0
1
0
0
Open post
sebffischer
sebffischer @sebffischer@mastodon.social · May 01, 2026
sebffischer
@sebffischer@mastodon.social

PhD Student in Statistics / Machine Learning.
Working on mlr3, which is a machine learning framework in R. Also working on a code transformation framework for R (jit compilation, autodiff): https://r-xla.github.io/anvl/

mastodon.social
Replying to @sebffischer@mastodon.social
Generally one pain is the:" select i-th axis, where i is a variable", because you can't do x[,,,2] because you don't know how many of those ","s you need.
0
1
0
0
Open post
sebffischer
sebffischer @sebffischer@mastodon.social · May 01, 2026
sebffischer
@sebffischer@mastodon.social

PhD Student in Statistics / Machine Learning.
Working on mlr3, which is a machine learning framework in R. Also working on a code transformation framework for R (jit compilation, autodiff): https://r-xla.github.io/anvl/

mastodon.social

What's the best "data.table"-like DSL for n-dimensional arrays? I am already aware of DumPy: https://dynomight.net/dumpy/

I am exploring ideas how to make array computations more readable. #rstats

0
2
1
0
Open post
sebffischer
sebffischer @sebffischer@mastodon.social · May 01, 2026
sebffischer
@sebffischer@mastodon.social

PhD Student in Statistics / Machine Learning.
Working on mlr3, which is a machine learning framework in R. Also working on a code transformation framework for R (jit compilation, autodiff): https://r-xla.github.io/anvl/

mastodon.social
Replying to @dodecadron@datasci.social
@dodecadron@datasci.social the capacity of a vector is the true length of a vector. Like I might know that I have to allocate up until n elements, so I could do> x <- with_capacity("integer", n) length(x) #> 0 x[n] <- 100 # no reallocation because internal length is already n length(x) #> n
1
1
0
0
Open post
sebffischer
sebffischer @sebffischer@mastodon.social · May 01, 2026
sebffischer
@sebffischer@mastodon.social

PhD Student in Statistics / Machine Learning.
Working on mlr3, which is a machine learning framework in R. Also working on a code transformation framework for R (jit compilation, autodiff): https://r-xla.github.io/anvl/

mastodon.social

Why do I have to write C Code in R to pre-allocate a vector with a certain capacity? #rstats

mastodon.social

Mastodon

0
2
0
0
Open post
sebffischer
sebffischer @sebffischer@mastodon.social · Apr 27, 2026
sebffischer
@sebffischer@mastodon.social

PhD Student in Statistics / Machine Learning.
Working on mlr3, which is a machine learning framework in R. Also working on a code transformation framework for R (jit compilation, autodiff): https://r-xla.github.io/anvl/

mastodon.social

Big new release in {anvl} (formerly called anvil, which conflicted with the AnVIL bioconductor package): https://github.com/r-xla/anvl/releases/tag/v0.2.0

There are too many features to list here, but one of the biggest ones is the introduction of eager mode, so usability of the package should be greatly improved!

#rstats

0
0
2
0
Open post
sebffischer
sebffischer @sebffischer@mastodon.social · Apr 21, 2026
sebffischer
@sebffischer@mastodon.social

PhD Student in Statistics / Machine Learning.
Working on mlr3, which is a machine learning framework in R. Also working on a code transformation framework for R (jit compilation, autodiff): https://r-xla.github.io/anvl/

mastodon.social

Has someone managed to publish a package on CRAN that downloads some prebuilt binaries? The C++ library I depend on uses LLVM and MLIR and the build is pretty complicated and I would really not like to maintain it as part of the CRAN package. #rstats

mastodon.social

Mastodon

1
2
6
0
Open post
sebffischer
sebffischer @sebffischer@mastodon.social · Apr 20, 2026
sebffischer
@sebffischer@mastodon.social

PhD Student in Statistics / Machine Learning.
Working on mlr3, which is a machine learning framework in R. Also working on a code transformation framework for R (jit compilation, autodiff): https://r-xla.github.io/anvl/

mastodon.social
Replying to @sebffischer@mastodon.social
@bioinfhotep@genomic.social did you manage to set everything up?
0
1
0
0
Open post
sebffischer
sebffischer @sebffischer@mastodon.social · Apr 19, 2026
sebffischer
@sebffischer@mastodon.social

PhD Student in Statistics / Machine Learning.
Working on mlr3, which is a machine learning framework in R. Also working on a code transformation framework for R (jit compilation, autodiff): https://r-xla.github.io/anvl/

mastodon.social
Replying to @bioinfhotep@genomic.social
@bioinfhotep@genomic.social anvil won't work on TPUs for now. Also (answering on a different thread): I update the docker docs on how to use GPUs
1
2
0
0
Open post
sebffischer
sebffischer @sebffischer@mastodon.social · Apr 19, 2026
sebffischer
@sebffischer@mastodon.social

PhD Student in Statistics / Machine Learning.
Working on mlr3, which is a machine learning framework in R. Also working on a code transformation framework for R (jit compilation, autodiff): https://r-xla.github.io/anvl/

mastodon.social
Replying to @bioinfhotep@genomic.social
@bioinfhotep@genomic.social I think the image should hopefully be smaller soon when we are using cuda12.8 package also for the container and just use the nvidia base image containing the drivers
1
0
0
0
Open post
sebffischer
sebffischer @sebffischer@mastodon.social · Apr 19, 2026
sebffischer
@sebffischer@mastodon.social

PhD Student in Statistics / Machine Learning.
Working on mlr3, which is a machine learning framework in R. Also working on a code transformation framework for R (jit compilation, autodiff): https://r-xla.github.io/anvl/

mastodon.social
Replying to @sebffischer@mastodon.social
@bioinfhotep@genomic.social Also I would use the "latest" image as a lot was added since the first release
0
2
0
0
Open post
sebffischer
sebffischer @sebffischer@mastodon.social · Apr 19, 2026
sebffischer
@sebffischer@mastodon.social

PhD Student in Statistics / Machine Learning.
Working on mlr3, which is a machine learning framework in R. Also working on a code transformation framework for R (jit compilation, autodiff): https://r-xla.github.io/anvl/

mastodon.social
Replying to @sebffischer@mastodon.social
@bioinfhotep@genomic.social but yeah, from your screenshot it's clear why the container is not working. You need the nvidia container toolkit to give docker GPU access. This is nothing that can be fixed by the image itself.
1
1
0
0
Open post
sebffischer
sebffischer @sebffischer@mastodon.social · Apr 19, 2026
sebffischer
@sebffischer@mastodon.social

PhD Student in Statistics / Machine Learning.
Working on mlr3, which is a machine learning framework in R. Also working on a code transformation framework for R (jit compilation, autodiff): https://r-xla.github.io/anvl/

mastodon.social
Replying to @sebffischer@mastodon.social
@bioinfhotep@genomic.social I wonder why the container does not work though ...
0
2
0
0
Open post
sebffischer
sebffischer @sebffischer@mastodon.social · Apr 19, 2026
sebffischer
@sebffischer@mastodon.social

PhD Student in Statistics / Machine Learning.
Working on mlr3, which is a machine learning framework in R. Also working on a code transformation framework for R (jit compilation, autodiff): https://r-xla.github.io/anvl/

mastodon.social
Replying to @bioinfhotep@genomic.social
@bioinfhotep@genomic.social manually setting up cuda dependencies is always a pain. But with the cuda12.8 package you basically only need a compatible driver.
1
3
0
0
Open post
sebffischer
sebffischer @sebffischer@mastodon.social · Apr 19, 2026
sebffischer
@sebffischer@mastodon.social

PhD Student in Statistics / Machine Learning.
Working on mlr3, which is a machine learning framework in R. Also working on a code transformation framework for R (jit compilation, autodiff): https://r-xla.github.io/anvl/

mastodon.social
Replying to @bioinfhotep@genomic.social
@bioinfhotep@genomic.social How did you install in WSL2? Did you use the cuda12.8 package?
0
9
0
0
Open post
sebffischer
sebffischer @sebffischer@mastodon.social · Apr 19, 2026
sebffischer
@sebffischer@mastodon.social

PhD Student in Statistics / Machine Learning.
Working on mlr3, which is a machine learning framework in R. Also working on a code transformation framework for R (jit compilation, autodiff): https://r-xla.github.io/anvl/

mastodon.social
Replying to @bioinfhotep@genomic.social
@bioinfhotep@genomic.social there is this docker image: https://github.com/r-xla/docker/blob/main/cuda-base/Dockerfile not sure whether it will work in colab
1
14
1
0
Open post
sebffischer
sebffischer @sebffischer@mastodon.social · Apr 18, 2026
sebffischer
@sebffischer@mastodon.social

PhD Student in Statistics / Machine Learning.
Working on mlr3, which is a machine learning framework in R. Also working on a code transformation framework for R (jit compilation, autodiff): https://r-xla.github.io/anvl/

mastodon.social
Replying to @sebffischer@mastodon.social
The main advantage of this package would be high customization ability + execution on accelerators. Also CPU execution should be pretty good due to the jit compilation I would expect.
1
0
0
0
Open post
sebffischer
sebffischer @sebffischer@mastodon.social · Apr 18, 2026
sebffischer
@sebffischer@mastodon.social

PhD Student in Statistics / Machine Learning.
Working on mlr3, which is a machine learning framework in R. Also working on a code transformation framework for R (jit compilation, autodiff): https://r-xla.github.io/anvl/

mastodon.social

Would anyone be interested in implementing a bayesian package (NUTS, HMC, etc.) on top of https://github.com/r-xla/anvil? #rstats

4
17
4
0
Open post
sebffischer
sebffischer @sebffischer@mastodon.social · Mar 02, 2026
sebffischer
@sebffischer@mastodon.social

PhD Student in Statistics / Machine Learning.
Working on mlr3, which is a machine learning framework in R. Also working on a code transformation framework for R (jit compilation, autodiff): https://r-xla.github.io/anvl/

mastodon.social

A few weeks ago I released (so far only on r-universe) the 0.1.0 version of my array computing framework for R (jit compiles, supports autodiff and runs on CPU and GPU): Check it out here: https://github.com/r-xla/anvil #rstats

13
0
14
0
Open post
sebffischer
sebffischer @sebffischer@mastodon.social · Sep 22, 2025
sebffischer
@sebffischer@mastodon.social

PhD Student in Statistics / Machine Learning.
Working on mlr3, which is a machine learning framework in R. Also working on a code transformation framework for R (jit compilation, autodiff): https://r-xla.github.io/anvl/

mastodon.social
Replying to @sebffischer@mastodon.social
@dgkf@fosstodon.org While I agree that a static type checker would be amazing, it would be a much more ambitious project. Also, I have personally come to the realization (while using S7) that one huge benefit of the type annotations are not only the (static) type checking, but to make it easier for me, the developer, to reason about code, and I also get this with dynamic input checking.
0
1
0
0
Open post
sebffischer
sebffischer @sebffischer@mastodon.social · Sep 22, 2025
sebffischer
@sebffischer@mastodon.social

PhD Student in Statistics / Machine Learning.
Working on mlr3, which is a machine learning framework in R. Also working on a code transformation framework for R (jit compilation, autodiff): https://r-xla.github.io/anvl/

mastodon.social
Replying to @dgkf@fosstodon.org
@dgkf@fosstodon.org I have also created an issue in S7 with a feature request for "S7 functions": https://github.com/RConsortium/S7/issues/564 where the idea is to extend S7 to also include typed functions, where the types are converted into a stopifnot(inherits(...)) block.
1
1
1
0
Open post
sebffischer
sebffischer @sebffischer@mastodon.social · Oct 09, 2024
sebffischer
@sebffischer@mastodon.social

PhD Student in Statistics / Machine Learning.
Working on mlr3, which is a machine learning framework in R. Also working on a code transformation framework for R (jit compilation, autodiff): https://r-xla.github.io/anvl/

mastodon.social
Replying to @jrosell@mastodon.social
@jrosell @elianhugh arguably, some of the confusion in r might also come from the lack of scalars. I think it makes sense that x[,1] returns a vector if x is a matrix and 1 was a scalar but x[, c(1)] returns a 2d matrix with the second dimension being 1 (here c(1) denotes a vector length 1).
0
0
0
0

Remote instance

mastodon.social
Open on original server
313k7r1n3
Elektrine

Tor hidden service

elekhj7afj4qnrr4yd3bkzslsyo5jgfxw3orgjkhlcxifueodybyiiad.onion

Platform

  • Email
  • Chat
  • Timeline
  • Communities
  • VPN
  • DNS

Company

  • About
  • Contact
  • FAQ

Legal

  • Terms of Service
  • Privacy Policy
  • Warrant Canary
  • Lite (no JS)
  • VPN Policy
  • Source code

Support

  • support@elektrine.com
  • Report Security Issue
Mail client setup IMAP mail.elektrine.com:993 POP3 mail.elektrine.com:995 SMTP mail.elektrine.com:465
© 2026 Elektrine. All rights reserved. Server: 00:00:22 UTC