Worked on: @reactjs, @messenger, @reasonml, @rescriptlang Currently: @midjourney, Pretext, Freerange, Handwired
California, USA
Joined January 2015
- Tweets8.8K
- Following521
- Followers78.8K
- Likes7.2K
Cheng Lou retweeted
I haven't been this excited about smartphone apps since 2008.
The iPhone Duo seems to have galvanized the developer community, leading to an explosion of novel concepts and new designs. Feels like the iPhone 3G all over again, and I'm *very much* here for it
Cheng Lou retweeted
MiniMax M3.1 Flash: Space Performance + Bunny Speed
🚀🐰 Yes, Space Bunny - which has topped global leaderboards with 200T+ tokens - was an earlier version of M3.1 Flash.
💰Redefining the performance economics of agentic AI. M3.1 Flash achieves frontier-level intelligence across key tasks such as Cowork (Automation-Bench: 62%; ALE: 30%), Coding (DeepSWE: 74%) and Visual Reasoning (BabyVision: 91%), at 1/3 the cost of M3 and 3x the speed.
API pricing (flat rate): $0.15 / $0.60 per 1M input/output tokens, with cache reads at $0.015 / 1M tokens.
Model weights will arrive on Oct 23.
API: platform.minimax.io
M Plan: minimax.io/m-plan
MiniMax Code for desktop: agent.minimax.io/download
Cheng Lou retweeted
Radical Pie uses Slug to draw its entire UI as vector graphics on the GPU. Not just text. Buttons, icons, check boxes, scroll bars, sliders, palettes, ..., everything.
Cheng Lou retweeted
Made a thing.... An ASCII sweater made from the lyrics of Weezer’s “Undone (The Sweater Song).” Grab a thread, give it a pull, and watch the whole sweater unravel word by word, line by line.
Part ode to Weezer, part ode to @_chenglou ’s Pretext library.
try it -> undone-record.vercel.app/
#ThreeJS #CreativeCoding
Ok hear me out, upcoming Macs are rumored to have a 512GB RAM (!) option. What if we just made a machine that ditched main memory? It'd be a herculean effort, plus making RAM non-volatile would suck in many ways, but fixable through AI at writing time or runtime, maybe better suited for a new generation of OSes
The giant upsides are tremendous low-level kernel + software simplifications, no more disk roundtrip, 100x throughput without caching, etc? Imagine ditching file system in favor of pure in-mem data (still standardizable at the boundary) then spill the rest to cloud like today
Cheng Lou retweeted
Hello Software Engineers, I lead TypeSafe AI Model Capabilities team. We are hiring. Here are a few qualities we are looking for:
• strong plus for people who resonate with the title "Experience Engineers": mobile engineers, full stack engineers, or even (virtual) reality engineers.
• strong coder (you need to be able to manually write code)
• product oriented thinker
I promise are fun! I trust that you will figure out how to apply.
Cheng Lou retweeted
✨ curious about how it's made? read along!
huge thanks @_chenglou for handwired
karanganesan.com/pet/blog
Cheng Lou retweeted
i made the new Roadster land with its thrusters… by turning JavaScript code into a neural net 🚀
used Handwired and gradient
descent to figure out when to fire the thrusters and stick the landing
all running in the browser
Cheng Lou retweeted
Made a tool that learns spring animation settings from your mouse movements. Using handwired by @_chenglou
teach-a-spring.vercel.app
Cheng Lou retweeted
wanted to understand @_chenglou’s Handwired a little better, so i built a tiny rocket that learns to land 🚀
the entire simulation is a neural net written in TypeScript. ran it backwards with gradients to find the thrust inputs that land it
no pretrained models. all in the browser
Cheng Lou retweeted
saw this and made meloform: sing a note, turn your voice into an instrument ♡
handwired ’s backprop fits the frequency components of tiny voice snippets, then a synth rebuilds them at different pitches
our voices can now be a piano with 88 keys!
link & github below 👇
Cheng Lou retweeted
Played around with @_chenglou new Handwired project last night
First I had to wrap my head around the implications…and I found this idea of adaptive UIs really intriguing
This example is a re-forming tree based on tap targets
What happens in we unfreeze some of these weights and actually train them? Well you can apply existing neural net techniques to it, losses and backprops and all, and now the UI becomes a bit more “learned”
Here’s a photo gallery where on the left I edit the rectangles and the right side gives me a new view where the photos learn to align themselves, training in real time as you drag (because the param count is so tiny).
This demo done fully in JavaScript, with JS looping through them numbers and backpropping. Stupid but fun! I’ve exposed the loss param slider you can play around with
Cheng Lou retweeted
This thing is insane. I ran a web app backwards by turning it into a neural net and doing gradient descent.
This is a calculator with handwriting input. It looks like a web app, but it’s actually one neural net running every frame, rendering the app and recognizing handwriting simultaneously.
How? I built the basic calculator in JS, compiled it to an NN with Handwired, then spliced a handwriting recognition model directly into the weights.
Even more cursed: since the whole app is an NN, we can run it in reverse! Start with the answer, then gradient descent flows backwards through both handwriting and app logic, and finds inputs that produce that answer. WTF
Thank you @_chenglou, this is so much fun!