Henry Hale

💻 building something awesome

[Henry Hale @devhenryhale]

Thanks to @JS_ORG

XTerminal docs are now on https://t.co/63FHqoQelg 🚀

[Henry Hale @devhenryhale]

yes, the beauty is in the model architecture. suppose that model X has 3B params and Y has 100B.
when both are trained on the similar data, they could both perform well. the diff is in the space used to capture and hold the information. Y (sparse) uses more than X(compact).

[orcus108 @orcus108]

WHAT THE HELL is happening in AI?

A 3B parameter model just put up coding benchmark scores in the same league as Claude Opus 4.5.

3 BILLION.

The weights are on Hugging Face, anyone can test it.

I genuinely don't know if this is a breakthrough or if the benchmarks are broken.

[Andrej Karpathy @karpathy]

- Drafted a blog post
- Used an LLM to meticulously improve the argument over 4 hours.
- Wow, feeling great, it’s so convincing!
- Fun idea let’s ask it to argue the opposite.
- LLM demolishes the entire argument and convinces me that the opposite is in fact true.
- lol

The LLMs may elicit an opinion when asked but are extremely competent in arguing almost any direction. This is actually super useful as a tool for forming your own opinions, just make sure to ask different directions and be careful with the sycophancy.

[Xenova @xenovacom]

IBM just released Granite 4.0 1B Speech, a compact and efficient speech-language model, designed for multilingual speech recognition and bidirectional speech translation. New #1 on the OpenASR leaderboard!

It can even run in your browser on WebGPU, thanks to 🤗 Transformers.js

[Henry Hale @devhenryhale]

@_trish_xD https://t.co/GdftVwCHLq

[Henry Hale @devhenryhale]

i built a todo list app and integrated the ai/ml model - the live demo uses onnx web runtime.

when setup locally, you can also run a local server that exposes the model via a Flask API.

[Henry Hale @devhenryhale]

Voice is cool, but text is still the fastest way to think, command, and build with AI.

Some thoughts on UX, speed, and preference for most interactions with AI 👇
https://t.co/7WJhxMz0LS

[Henry Hale @devhenryhale]

https://t.co/xsivIwuqhG

[Henry Hale @devhenryhale]

before you upload that file to a server-based tool (for background removal, audio/video conversion, image-to-text OCR), try i💙Media - a collection of powerful media utilities that run entirely in your browser.
👉 https://t.co/l7pJaa1pgv

[Henry Hale @devhenryhale]

worstcase is powered by @babeljs (such an amazing project)

JSX & TSX supported too

[Henry Hale @devhenryhale]

🔥 Just shipped: worstcase - a javascript complexity analyzer!

Automatically detects O(n²) loops, memory leaks, and performance bottlenecks in your JS code - no pattern matching, pure AST analysis.

📦 npm install worstcase

https://t.co/VqYTkUbvJp

[Henry Hale @devhenryhale]

🚀 Just released "JavaScript Complexity Analyzer" for VS Code!

See Big O complexity directly in your editor with inline decorations. Supports JS, TS, JSX, and TSX files.

✨ Real-time analysis
⚡ Zero config setup
🆓 Completely free

https://t.co/0VFImbNl62

[Henry Hale @devhenryhale]

@cory_laviska @shadcn @webawesomer @shoelace_style not a shadcn theme but rather every ui component packaged as a web component that can be lazy-loaded. just like how shoelace/webawesome works.

the shift from an open-source & free shoelace library to webawesome with paid components made it hard to build anything on top of it.

[Henry Hale @devhenryhale]

We need @shadcn ui as a web components library 🙏

The design system is incredible, but being React-only limits its reach. Imagine those beautiful, accessible components working across vanilla JS, Vue, Svelte, Angular...

[Henry Hale @devhenryhale]

had a side project noted for prototyping a while back.
>essentially a no-code code platform
when and why did @shoelace_style become @webawesomer?

[Henry Hale @devhenryhale]

@samwhoo well then check this out:
https://t.co/a2dzrJfxi0

[Henry Hale @devhenryhale]

Nice post! Been building an automated Big O complexity analyzer for JavaScript code.
>uses pattern matching to detect common complexity patterns.
>built a demo with Monaco Editor integration where you can paste JS code and get real-time complexity analysis for each function.

[Sam Rose @samwhoo]

✨ Big O ✨

Let me take you on a visual introduction to what big O notation is in my new blog post: https://t.co/Nt1jDSRTY8.

With big O notation you can better understand how algorithms will perform, finding orders of magnitude improvements with very simple changes to your code

[Henry Hale @devhenryhale]

live demo with real-time complexity analysis as you type
👇
https://t.co/a2dzrJfxi0

[Henry Hale @devhenryhale]

when the javascript program grows(1k+ lines), the editor is a mess with long & many complexity results.

so i add an option to follow standard protocols - drop all coefficients. O(2n) becomes O(n) and O(32) = O(1)

[Henry Hale @devhenryhale]

Try it yourself: https://t.co/akC5moccbz

[Henry Hale @devhenryhale]

🧐 What if your editor could warn you about O(n²) algorithms in real-time?

Built exactly that. No hardcoded patterns - pure code structure analysis.

30-second demo showing nested loops, recursion, and built-in method complexity 👇

[Henry Hale @devhenryhale]

🔥 Just shipped: worstcase - a javascript complexity analyzer!

Automatically detects O(n²) loops, memory leaks, and performance bottlenecks in your JS code - no pattern matching, pure AST analysis.

📦 npm install worstcase

https://t.co/VqYTkUbvJp