Петр Умстер

Активный серфер в интернет. Веду активный здоровый образ жизни. Увлекаюсь экологическим строительством. Работаю Веб-программистом.

[Taelin @VictorTaelin]

Status update: I've been on/off AI agents in the last few days and it is a verifiable truth that every day I didn't use agents, I was more productive. I still attribute that to how slow they are, and my own inability to multi-task efficiently. The magic is there but the slowness doesn't let it cross the threshold where they actually make me faster, and I still dislike the whole thinking paradigm.

About Bend2: honestly, the C/Metal compiler codebase is a clusterfuck right now. I regret letting AI agents write it. All tests pass, and GPU performance is mind-blowing, so the core architecture works. Yet, it has a LOT of bugs. Anything not covered by the tests is a coin toss. This is actually impressive, because, in many parts of the codebase, the right solution was actually the simplest one, yet, the agents STILL managed to find a way to make it work just for the tests. The level of reward hack these agents output is actually impressive I can't even be mad.

It is also ironical because that's the very problem that Bend's proof system was supposed to solve, but Bend is in TypeScript, not in Bend. I'm disappointed I didn't write Bend in itself, and now I feel an immense urge to do so. But the clock is ticking . . .

Still, I do not think Bend is worth launching without the GPU compiler being solid, because the closest competitor, Lean, is actually extremely good, so we need a big differential. Yet, due to the very nature of the project, it would be embarrassing to have bugs at launch.

Regarding AI, I now believe using current gen AI agents in production codebase is harmful and a massive mistake. That doesn't mean no agents at all, but agents work best when they don't touch critical code. Debugging, researching, providing insights, scripts / tools, or anything that doesn't touch code you will maintain in the long term. But if you merge AI code without reading, you're going to have a bad time. Speaking from experience

I'm working 10h/day on SupGen and the remaining time on Bend2

[Theo - t3.gg @theo]

@creatine_cycle Finally a banger

[Open Design @OpenDesignHQ]

📢📢📢Open Design 0.5.0 is live! 🚀

From “render once” to “iterate forever”:

-Live-data dashboard workflows
-Inspect mode for per-element tuning
-Accent theme control across launcher
-Critique Theater Phase 5
-Qoder CLI + Nano Banana
-Linux headless lifecycle

Built fast with the community: 51 PRs, 36 contributors, ~1 day.
Try it now👉https://t.co/aPlY6id6dl

[spidey @lochan_twt]

The day a blind man sees. The first thing he throws away is the stick that has helped him all his life

[Arnav Gupta @championswimmer]

https://t.co/BwpF1bXhIr

[Zed @zeddotdev]

We've shipped more than a thousand versions of Zed, but all of them began with zero. Today, that changes.

https://t.co/AJ0crNOFhU

[Петр Умстер @umster_ru]

@nahcrof I’m extremely grateful for everything you do!

[Петр Умстер @umster_ru]

@jullerino @tanavtwt Could you please describe in more detail how to do this?

[Петр Умстер @umster_ru]

@jullerino @r_marked @t3dotchat Is it possible to do this in the Windows version, so that I can choose where the program should be installed?

[Aakash Gupta @aakashgupta]

Karpathy told Dwarkesh that a 1 billion parameter model, trained on clean data, could hit the intelligence of today's 1.8 trillion parameter frontier.

That is a 1,800x compression claim. The math behind it is more defensible than it sounds.

When researchers at frontier labs look at random samples from their training corpus, they see stock ticker symbols, broken HTML, forum spam, autogenerated gibberish. Not Wikipedia. Not the Wall Street Journal. The actual pretraining dataset is mostly noise, and the model is burning parameters to vaguely remember all of it.

One estimate pegs Llama 3's information compression at 0.07 bits per token. Well-structured English carries around 1.5 bits per token of real information. The trillion-parameter model is holding a roughly 5% resolution image of the internet it trained on.

So when a lab ships a 1.8 trillion parameter model, the overwhelming majority of those weights are handling rough memorization. They are compression overhead for a noisy training set, taking up capacity that could be doing reasoning instead.

Karpathy's proposal is to separate the two. Build a cognitive core: a small model that contains only the algorithms for reasoning and problem-solving, stripped of encyclopedic memorization. Pair it with external memory the model queries when it needs a fact. A 1 billion parameter reasoner plus retrieval beats a 1.8 trillion parameter model trying to do both.

The data already supports this direction. GPT-4o runs at roughly 200 billion parameters and outperforms the original 1.8 trillion GPT-4. Inference costs for GPT-3.5 level performance fell 280x between 2022 and 2024, driven almost entirely by smaller, cleaner, better-architected models. The trend line is pointing where Karpathy says it should.

The real implication for anyone tracking the AI trade: data quality is the actual constraint. The companies winning the next phase will be the ones who figured out what to train on, and what to throw away.

[Uncle Bob Martin @unclebobmartin]

What we are losing with AI is syntax -- and good riddance. The less our brains are occupied by semicolons and braces the better. There are much more important things for us to consider and manage.

[Guille Martín @Farmaenfurecida]

A veces es necesario:

[Xiaomi MiMo @XiaomiMiMo]

Xiaomi MiMo Token Plan is here.

One subscription. All modalities.
Build with MiMo-V2-Pro, Omni, and TTS.

No 5-hour limits. No throttling.
Transparent usage and billing. Just ship.

Works with whatever you use: @openclaw, @opencode, @kilocode, @cline, @roocode
Includes priority beta access to our newest models.

12% off your first purchase. TTS is free for now.
Subscribe now → https://t.co/MJ8GAIJGPW

[Dane Knecht 🦭 @dok2001]

Announcing: EmDash, the WordPress spiritual successor built for the modern web.

TypeScript. Serverless. MIT licensed. x402 for agent-era monetization. MCP server built in. Deploy to Cloudflare or anywhere Node.js runs.

Imports your existing WordPress site in minutes.

npm create emdash@latest
https://t.co/vgqc4tQn30

[Quinn Slack @sqs]

Crazy times

https://t.co/KfMJrbRGs7

[GitHub Projects Community @GithubProjects]

∴｡
　･ﾟ*｡.
　　･ *ﾟ
　 ･ ﾟ*
･｡
*･｡
*.｡
｡･
°*. ﾟ｡
°*.
ﾟ*.
｡｡ ･
｡ ･ﾟ
｡°*.
｡*･｡

[Петр Умстер @umster_ru]

@mattlam_ Will the web version be similar to pi-gui? We're really looking forward to an alternative to opencode web.

[Matthew Lam @mattlam_]

pi-gui takes another step to provide a codex app vibe/ux for pi fans.

Build now includes tool expansion/collapse, in timeline diff viewer, diff reviewer panel, @ file references. Hope you like it!

Also works off the bat with my new pi subagents extension, and should for other extension tools too :) Will explore better extensions support for the gui experience too.

[Oliver Prompts @oliviscusAI]

someone built a web-based System Design Simulator.

you drag and drop components (api gateways, dbs, caches) and it actually simulates real-time traffic.

you can watch latency, bottlenecks, and failures happen live...

[Cerebras @cerebras]

Nvidia DLSS 5 off DLSS 5 on