#Walerie

12.9K posts

#Walerie banner
#Walerie

#Walerie

@ForecastFire

certafide & bonefied. Creator of DistinctionBench, testing for representation-invariant structure abstraction. previously ForecastFire

Katılım Aralık 2017
7.5K Takip Edilen541 Takipçiler
Sabitlenmiş Tweet
#Walerie
#Walerie@ForecastFire·
This is the ideal realm of forms. You may not like it, but this is what peak abstraction looks like.
#Walerie tweet media
#Walerie@ForecastFire

@voooooogel From preliminary results I suspect this is also a test of big model smell

English
4
3
26
2.3K
#Walerie
#Walerie@ForecastFire·
ohhh yeah. containers / depth value hits way different than place value, cuz it's embodied. While I got ur attention, I'm working on a tool/puzzle game to help develop more of these embodied/iconic number intuitions -- like an abacus but for k-12 algebra. would you be down for a playtest sometime? I recently had a breakthrough on the interaction model so I expect to have some nice developments on it end of this month!
English
0
0
1
15
Andy Masley
Andy Masley@AndyMasley·
I have claude code automatically uploading every substack post I make to a really luxurious page on my website now, only took a few prompts to make this happen andymasley.com/writing/a-chea…
English
4
1
74
3.6K
#Walerie
#Walerie@ForecastFire·
@AndyMasley Oh right I already tweeted this graph at you lol
English
0
0
1
10
#Walerie
#Walerie@ForecastFire·
@AndyMasley THE PEOPLE CRAVE ICONIC ARITHMETIC.... yesss we need more unary tallies and groupings
#Walerie tweet media
English
2
0
3
70
#Walerie retweetledi
Andrew Jefferson
Andrew Jefferson@EastlondonDev·
This REPL isn’t like anything you’ve seen before. It’s running on a neural network. It’s a feed forward network, using attention, and it implements a fully working wasm interpreter. When I saw an article on the topic of wasm interpreting llms I had to build this. Wasm + the tech behind AI + a repl … running in a browser and built and deployed on Replit - this is an homage to @amasad, one of my personal heroes 🐐 The idea is based on a recent post by @PerceptaAI but they didn’t provide the model architecture or really any of the important implementation details so I had to do a lot of figuring out and testing with the Replit agent to build it - it’s mind blowing that AI can produce this - a neural network architecture with hand crafted weights to implement something too new to be in any of its training data. P.s. it was all done on my phone, most of it while flying from SFO to Munich, what a time to be alive!
Andrew Jefferson tweet media
English
19
22
197
74.4K
#Walerie retweetledi
Jonny Miller
Jonny Miller@jonnym1ller·
I think what @attncopilot are doing is super interesting. Essentially, they're teaching unclenching practice whilst you're at work to find more flow + enjoyment. (i.e. its a superb inner-work trojan horse + typically leads to more productivity)
Jonny Miller tweet media
English
0
6
111
4.8K
#Walerie retweetledi
Anthropic
Anthropic@AnthropicAI·
We invited Claude users to share how they use AI, what they dream it could make possible, and what they fear it might do. Nearly 81,000 people responded in one week—the largest qualitative study of its kind. Read more: anthropic.com/features/81k-i…
English
332
884
6.1K
2.3M
#Walerie retweetledi
Andy Masley
Andy Masley@AndyMasley·
Something that's driven me insane in climate communication is that there hasn't been (to my knowledge) a carbon footprint calculator that compares your individual cuts to things you can do to green the grid. I vibecoded one and I'm pretty happy with it: andymasley.com/visuals/carbon…
Andy Masley tweet media
English
7
10
156
8.6K
Liminal Warmth ❤️‍🔥
Liminal Warmth ❤️‍🔥@liminal_warmth·
You could take this up a notch and randomize the language (and maybe grammar) assignment on each game so that it both has replayability and can't be easily solved by looking at a wiki, forcing you to actually play the primary game loop of language acquisition.
Indie Game Joe@IndieGameJoe

This indie dev is making a game where you must learn a complete alien language from scratch. - Talk to cute alien creatures to earn their trust - Decode an alien alphabet - Progress depends on what you understand It's called Pikku Adventure. Looks amazing!

English
6
1
139
4.5K
#Walerie retweetledi
Lorenzo Xiao
Lorenzo Xiao@lrzneedresearch·
I dont know who needs this but this is like an overview of notes I wrote to prepare for my "Agentic system design" Let me know if anyone feel like this is helpful and I can write more in details for each particular section algoroxyolo.github.io/blog/2026/llm-…
English
9
11
107
7K
#Walerie
#Walerie@ForecastFire·
@ember_arlynx @sameQCU I'm v curious what sort of context meter would be more cybernetic and of what inspiration But also FYI y'all r both in SF presently I think
English
1
0
3
20
cmr://ember
cmr://ember@ember_arlynx·
@sameQCU patched in a more cybernetic context meter myself
English
2
0
2
35
サメQCU
サメQCU@sameQCU·
claude code is missing multi account multi auth features (like you'd use to switch immediately between claude 4 work and claude 4 personal non work use by terminal aliases or a within app command) so i'm now using claude code to feature patch claude code
English
3
0
17
691
#Walerie retweetledi
davinci
davinci@leothecurious·
bennett is imo by far one of the brightest minds in cog sci today. he's almost single-handedly pushing the frontier of our understanding of consciousness with refreshingly grounded theories. this paper alone promises to resolve some technical gaps that often led to people assuming consciousness in arbitrary complex systems. very much enjoyed reading it.
Michael Timothy Bennett@MiTiBennett

English
36
186
2.4K
213.4K
#Walerie retweetledi
Zhikai Zhang
Zhikai Zhang@Zhikai273·
🎾Introducing LATENT: Learning Athletic Humanoid Tennis Skills from Imperfect Human Motion Data Dynamic movements, agile whole-body coordination, and rapid reactions. A step toward athletic humanoid sports skills. Project: zzk273.github.io/LATENT/ Code: github.com/GalaxyGeneralR…
English
163
642
4.1K
1.3M
#Walerie retweetledi
&.
&.@amplifiedamp·
someone should do this and find out what happens while logging the results online
&. tweet media
English
5
17
205
5.8K
#Walerie
#Walerie@ForecastFire·
@emollick i'm in the 5% ☺️ but also, thanks for sharing because I am working on this sort of problem rn.
English
0
0
1
72
Ethan Mollick
Ethan Mollick@emollick·
This is a really interesting post using the Enron email archive to test how good agents are at navigating work, and it provides some helpful evidence that agent swarms are less useful than agent organizations.
rohit@krishnanrohit

x.com/i/article/2032…

English
41
42
450
97.2K
#Walerie
#Walerie@ForecastFire·
@sameQCU just tested an api call opus 4.5, it worked
English
1
0
2
28
サメQCU
サメQCU@sameQCU·
is it just me or is claude api down (internal 500: type: error. error notation? "error.")
English
1
0
5
574
#Walerie retweetledi
Marianne
Marianne@thrialectics·
a hug is such a cybernetic gesture
English
2
6
30
730