Swirl.jl now supports multi-step code questions and full Markdown rendering! 🌀

1/2 Multi-step questions: you can now design exercises that require multiple code lines, with separate hints for each step. Perfect for walking learners through multi-part problems or short coding tasks.

🌀 Swirl.jl is here!
An interactive learning platform that teaches JuliaLang in the REPL.

✨ Features
🎓 Interactive code checks
📚 Structured lessons
💾 Auto-save
🎮 Reset anytime

For now, Julia Basics is the only course. It will be expanded and new courses are on the way.
github.com/atantos/Swir...

If you’re exploring linguistic change across Austen’s novels (1811–1817), you might want to ask questions like:
👉 Did the verb “might” become more associated with epistemic verbs after 1814?
👉 Which collocates changed most between 1811 and 1817?
This single line in JuliaLang does all the lifting 👇

Here is the mental model for TextAssociations.jl users:

Raw Text / Corpus
↓
ContingencyTable
↓
CorpusContingencyTable
↓
assoc_score(::Type{<:AssociationMetric}, x::AssociationDataFormat; kwargs...)
↓
Metric evaluator (eval_pmi, …)
↓
Result as DataFrame (or Vector if scores_only=true)

Here’s a 37-second sneak peek showing how the TextAssociations.jl package computes a Bayesian-smoothed version of PMI for the word "might" across Jane Austen’s novels, from text normalization and corpus loading to contingency-table construction and metric calculation.
#JuliaLang #CorpusLinguistics

In the attached snapshot, contingency tables are generated capturing words occurring within a 5-word window around "innovation" and PMI scores quantify their association strength.
The top 5 collocates are then displayed in a DataFrame. #TextAssociations #JuliaLang

Finally announcing Bonito.jl! 🎉 A Julia web framework that went from "neglected prototype" to something I'm genuinely proud of.

Reactive UIs, rich widgets, Makie integration, deploy anywhere (Jupyter/VSCode/web/static HTML)
makie.org/website/blog...

In language-related and digital humanities research, we are often concerned not only with meaning, but also with pattern and context — how words cluster within a discourse or register, across time periods, or within particular language varieties such as dialects and learner corpora. 1/n

I’ve created a GitHub repo where I’m translating the #Python code from @rasbt's Build a LLM from Scratch into #JuliaLang. I’d love for Julia enthusiasts to explore it and I’d be eager to see contributions and improvements. Don’t hesitate to ⭐ it! #LLMs #AI
github.com/atantos/Buil...

If you haven't found it already be sure to subscribe to the #JuliaDispatch podcast by @chrisrackauckas.bsky.social and Michael Tiemann! This week's special guest is Παναγιώτης Γεωργακόπουλος to discuss the #PlutoJL project.

youtu.be/TknY3rE1TeE?...

Continuing my journey of translating
@rasbt
's "Build a LLM (from Scratch)" Python code to Julia, my latest Medium article focuses on computing attention weights for all input tokens as discussed in section 3.3.2. #LLMs #Python #JuliaLang #Transformers #Attention #GenAI 1/2

Inspired by
@rasbt
Build a LLM (from Scratch), I’ve written a Medium article on implementing a simple self-attention mechanism using Python and Julia. It includes practical, theory-driven details to deepen understanding. #LLMs #Python #JuliaLang #Transformers #Attention #AI 1/2

Ok! Here we are! Blue clean skies ahead..

Just another day spent living life instead of posting about it...

How I missed these original twitter colors! #throwback