Frank Hull
@frankiethull.bsky.social
Here for R, MLOps, Energy, & local LLMs
FluidBench v1:
completion rate shows Opus 4.5 & Gemini 3 Pro leading the way.
completion rate shows Opus 4.5 & Gemini 3 Pro leading the way.
December 16, 2025 at 7:15 PM
FluidBench v1:
completion rate shows Opus 4.5 & Gemini 3 Pro leading the way.
completion rate shows Opus 4.5 & Gemini 3 Pro leading the way.
Reposted by Frank Hull
I have to say that it's very helpful to see the "tidymodels way" to build R packages using Python libraries. Thanks.
December 14, 2025 at 10:20 AM
I have to say that it's very helpful to see the "tidymodels way" to build R packages using Python libraries. Thanks.
Anyone binding chronos or moirai to R? 🧙🏻
Check it out!
What I thought was only a @topepo.bsky.social experiment, has now moved under tidymodels:
tabpfn.tidymodels.org
What I thought was only a @topepo.bsky.social experiment, has now moved under tidymodels:
tabpfn.tidymodels.org
Prior-Data Fitted Network Foundational Model for Tabular Data
R functions are used via Python to train and predict TabPFN models of Hollmann et al (2025) "Accurate predictions on small data with a tabular foundation model," Nature, 637(8045) <doi:10.1038/s41586-...
tabpfn.tidymodels.org
December 12, 2025 at 9:06 PM
Anyone binding chronos or moirai to R? 🧙🏻
Reposted by Frank Hull
Messing with this all day & just had a great out-of-sample prediction 🙌🏻
It works well if you split-apply-combine.
github.com/topepo/TabPFN
It works well if you split-apply-combine.
github.com/topepo/TabPFN
GitHub - topepo/TabPFN: Foundation Model for Tabular Data via reticulate
Foundation Model for Tabular Data via reticulate. Contribute to topepo/TabPFN development by creating an account on GitHub.
github.com
July 11, 2025 at 10:36 PM
Messing with this all day & just had a great out-of-sample prediction 🙌🏻
It works well if you split-apply-combine.
github.com/topepo/TabPFN
It works well if you split-apply-combine.
github.com/topepo/TabPFN
Check it out!
What I thought was only a @topepo.bsky.social experiment, has now moved under tidymodels:
tabpfn.tidymodels.org
What I thought was only a @topepo.bsky.social experiment, has now moved under tidymodels:
tabpfn.tidymodels.org
Prior-Data Fitted Network Foundational Model for Tabular Data
R functions are used via Python to train and predict TabPFN models of Hollmann et al (2025) "Accurate predictions on small data with a tabular foundation model," Nature, 637(8045) <doi:10.1038/s41586-...
tabpfn.tidymodels.org
December 12, 2025 at 8:29 PM
Check it out!
What I thought was only a @topepo.bsky.social experiment, has now moved under tidymodels:
tabpfn.tidymodels.org
What I thought was only a @topepo.bsky.social experiment, has now moved under tidymodels:
tabpfn.tidymodels.org
from shinywidgets import render_altair is all you need
December 11, 2025 at 4:09 PM
from shinywidgets import render_altair is all you need
Alrighty survived week 1, week 2 is underway.
Opening up the ol "Dashboards with Shiny for Python" quarto documentation🤞🏻
Didn't really factor in the fact a challenge may require creating a shiny app 😅
Opening up the ol "Dashboards with Shiny for Python" quarto documentation🤞🏻
Didn't really factor in the fact a challenge may require creating a shiny app 😅
This is the second year my team has put together an internal facing Advent Of Code!!
It is a data wrangling & tabular focused AOC.
The twist, this year my submissions will be in Python instead of R 🐍
It is a data wrangling & tabular focused AOC.
The twist, this year my submissions will be in Python instead of R 🐍
December 10, 2025 at 6:15 PM
Alrighty survived week 1, week 2 is underway.
Opening up the ol "Dashboards with Shiny for Python" quarto documentation🤞🏻
Didn't really factor in the fact a challenge may require creating a shiny app 😅
Opening up the ol "Dashboards with Shiny for Python" quarto documentation🤞🏻
Didn't really factor in the fact a challenge may require creating a shiny app 😅
devtools::check is all you need
December 10, 2025 at 5:32 PM
devtools::check is all you need
Reposted by Frank Hull
A new release of chores is now on #rstats CRAN!
There are now some LLMs tiny enough to run on a laptop that are capable of powering chores helpers. Here's a real-time demo with a model running on my Macbook.
Read more: www.simonpcouch.com/blog/2025-12...
There are now some LLMs tiny enough to run on a laptop that are capable of powering chores helpers. Here's a real-time demo with a model running on my Macbook.
Read more: www.simonpcouch.com/blog/2025-12...
December 10, 2025 at 3:25 PM
A new release of chores is now on #rstats CRAN!
There are now some LLMs tiny enough to run on a laptop that are capable of powering chores helpers. Here's a real-time demo with a model running on my Macbook.
Read more: www.simonpcouch.com/blog/2025-12...
There are now some LLMs tiny enough to run on a laptop that are capable of powering chores helpers. Here's a real-time demo with a model running on my Macbook.
Read more: www.simonpcouch.com/blog/2025-12...
I've learned a few acronyms to not use on this site 🤗
December 6, 2025 at 10:12 PM
I've learned a few acronyms to not use on this site 🤗
First year my main is a linux + nvidia laptop & I finally had to honor of purging & reinstalling a major GPU nvidia driver issue. 😬
I cannot figure this stuff out on my own at all.
While we critic everyday AI code, LLMs walked me through the whole process of the fix which was awesome.
I cannot figure this stuff out on my own at all.
While we critic everyday AI code, LLMs walked me through the whole process of the fix which was awesome.
December 5, 2025 at 4:32 PM
First year my main is a linux + nvidia laptop & I finally had to honor of purging & reinstalling a major GPU nvidia driver issue. 😬
I cannot figure this stuff out on my own at all.
While we critic everyday AI code, LLMs walked me through the whole process of the fix which was awesome.
I cannot figure this stuff out on my own at all.
While we critic everyday AI code, LLMs walked me through the whole process of the fix which was awesome.
Honored to have my code in here
top_k = "top_k"
top_k = "top_k"
December 3, 2025 at 5:52 PM
Honored to have my code in here
top_k = "top_k"
top_k = "top_k"
This is the second year my team has put together an internal facing Advent Of Code!!
It is a data wrangling & tabular focused AOC.
The twist, this year my submissions will be in Python instead of R 🐍
It is a data wrangling & tabular focused AOC.
The twist, this year my submissions will be in Python instead of R 🐍
December 3, 2025 at 4:30 PM
This is the second year my team has put together an internal facing Advent Of Code!!
It is a data wrangling & tabular focused AOC.
The twist, this year my submissions will be in Python instead of R 🐍
It is a data wrangling & tabular focused AOC.
The twist, this year my submissions will be in Python instead of R 🐍
stringr::str_glue is all you need
December 3, 2025 at 2:52 AM
stringr::str_glue is all you need
Reposted by Frank Hull
Add your voice to the R community! R!sk 2026 Call for Proposals is open through Dec 7, 2025. Share how you use R to calculate, measure, and mitigate risk across finance, insurance, climate, healthcare & more.
Submit now: rconsortium.github.io/Risk_website/
#rstats #riskmanagement
Submit now: rconsortium.github.io/Risk_website/
#rstats #riskmanagement
December 2, 2025 at 1:30 PM
Add your voice to the R community! R!sk 2026 Call for Proposals is open through Dec 7, 2025. Share how you use R to calculate, measure, and mitigate risk across finance, insurance, climate, healthcare & more.
Submit now: rconsortium.github.io/Risk_website/
#rstats #riskmanagement
Submit now: rconsortium.github.io/Risk_website/
#rstats #riskmanagement
Reposted by Frank Hull
duckdb-mlpack 0.0.5: Added kmeans, version helpers, documentation
Bringing mlpack machine learning to duckdb SQL
dirk.eddelbuettel.com/blog/2025/12...
#duckdb #datascience #mlpack
Added kmeans, version helpers, documention
Bringing mlpack machine learning to duckdb SQL
dirk.eddelbuettel.com/blog/2025/12...
#duckdb #datascience #mlpack
Added kmeans, version helpers, documention
December 2, 2025 at 6:45 PM
duckdb-mlpack 0.0.5: Added kmeans, version helpers, documentation
Bringing mlpack machine learning to duckdb SQL
dirk.eddelbuettel.com/blog/2025/12...
#duckdb #datascience #mlpack
Added kmeans, version helpers, documention
Bringing mlpack machine learning to duckdb SQL
dirk.eddelbuettel.com/blog/2025/12...
#duckdb #datascience #mlpack
Added kmeans, version helpers, documention
Reposted by Frank Hull
It's tomorrow!
This is next week! A fun discussion about ggpal2 & chores.
I think the coolest part is you should leave understanding how to create your own chore with little to no code.
If you're new to AI or want a low-barrier way to create an assistant with R, chores is a great place to start.
I think the coolest part is you should leave understanding how to create your own chore with little to no code.
If you're new to AI or want a low-barrier way to create an assistant with R, chores is a great place to start.
December 1, 2025 at 5:49 PM
It's tomorrow!
Tested a few new models for R packages & some small models on predicting numbers as lexicon tokens.
Gemini 3 Pro is a really good R programmer. Sonar is great for debugging, & gpt-oss is good at being a tense model. Ran Opus & Gemini side-by-side but preferred Gemini.
github.com/frankiethull...
Gemini 3 Pro is a really good R programmer. Sonar is great for debugging, & gpt-oss is good at being a tense model. Ran Opus & Gemini side-by-side but preferred Gemini.
github.com/frankiethull...
GitHub - frankiethull/tense: tense: LLM semantic sequence prediction for ordered data
tense: LLM semantic sequence prediction for ordered data - frankiethull/tense
github.com
November 29, 2025 at 4:49 PM
Tested a few new models for R packages & some small models on predicting numbers as lexicon tokens.
Gemini 3 Pro is a really good R programmer. Sonar is great for debugging, & gpt-oss is good at being a tense model. Ran Opus & Gemini side-by-side but preferred Gemini.
github.com/frankiethull...
Gemini 3 Pro is a really good R programmer. Sonar is great for debugging, & gpt-oss is good at being a tense model. Ran Opus & Gemini side-by-side but preferred Gemini.
github.com/frankiethull...
Reposted by Frank Hull
I spent a bit of time this weekend updating my #Quarto listing for talks and workshops 🤗
Check it out here: ivelasq.rbind.io/talk
Many thanks to Meghan Hall for sharing the template and CSS code! 💜
Check it out here: ivelasq.rbind.io/talk
Many thanks to Meghan Hall for sharing the template and CSS code! 💜
November 24, 2025 at 8:18 PM
I spent a bit of time this weekend updating my #Quarto listing for talks and workshops 🤗
Check it out here: ivelasq.rbind.io/talk
Many thanks to Meghan Hall for sharing the template and CSS code! 💜
Check it out here: ivelasq.rbind.io/talk
Many thanks to Meghan Hall for sharing the template and CSS code! 💜
This is next week! A fun discussion about ggpal2 & chores.
I think the coolest part is you should leave understanding how to create your own chore with little to no code.
If you're new to AI or want a low-barrier way to create an assistant with R, chores is a great place to start.
I think the coolest part is you should leave understanding how to create your own chore with little to no code.
If you're new to AI or want a low-barrier way to create an assistant with R, chores is a great place to start.
November 24, 2025 at 5:42 PM
This is next week! A fun discussion about ggpal2 & chores.
I think the coolest part is you should leave understanding how to create your own chore with little to no code.
If you're new to AI or want a low-barrier way to create an assistant with R, chores is a great place to start.
I think the coolest part is you should leave understanding how to create your own chore with little to no code.
If you're new to AI or want a low-barrier way to create an assistant with R, chores is a great place to start.
November 22, 2025 at 10:33 PM
Reposted by Frank Hull
📣🚨 Very much looking forward to hearing from @frankiethull.bsky.social about the intersection of coded data viz and AI assistants! 👀🚀
Join us Tuesday Dec 2nd, at 3pm Eastern to hear about {ggpal2}, a {chores} extension. #ggplot2 #ai #chores
bit.ly/RxAI-join
Join us Tuesday Dec 2nd, at 3pm Eastern to hear about {ggpal2}, a {chores} extension. #ggplot2 #ai #chores
bit.ly/RxAI-join
November 21, 2025 at 9:54 PM
📣🚨 Very much looking forward to hearing from @frankiethull.bsky.social about the intersection of coded data viz and AI assistants! 👀🚀
Join us Tuesday Dec 2nd, at 3pm Eastern to hear about {ggpal2}, a {chores} extension. #ggplot2 #ai #chores
bit.ly/RxAI-join
Join us Tuesday Dec 2nd, at 3pm Eastern to hear about {ggpal2}, a {chores} extension. #ggplot2 #ai #chores
bit.ly/RxAI-join
Reposted by Frank Hull
#TodayinHistory #dataviz #OTD 📊
📅Nov 20, 1869 Charles Joseph Minard published his iconic flow map graphic of Napoleon's March on Moscow.
Some called it best graphic ever produced, but EJ Marey put it best:
"Defies the pen of the historian in its brutal eloquence"
No known portrait of CJ, only .sig
📅Nov 20, 1869 Charles Joseph Minard published his iconic flow map graphic of Napoleon's March on Moscow.
Some called it best graphic ever produced, but EJ Marey put it best:
"Defies the pen of the historian in its brutal eloquence"
No known portrait of CJ, only .sig
November 21, 2025 at 1:18 AM
#TodayinHistory #dataviz #OTD 📊
📅Nov 20, 1869 Charles Joseph Minard published his iconic flow map graphic of Napoleon's March on Moscow.
Some called it best graphic ever produced, but EJ Marey put it best:
"Defies the pen of the historian in its brutal eloquence"
No known portrait of CJ, only .sig
📅Nov 20, 1869 Charles Joseph Minard published his iconic flow map graphic of Napoleon's March on Moscow.
Some called it best graphic ever produced, but EJ Marey put it best:
"Defies the pen of the historian in its brutal eloquence"
No known portrait of CJ, only .sig
tidyr::nest is all you need
November 17, 2025 at 7:52 PM
tidyr::nest is all you need