Anton Lobach (@AntLobach) header
Anton Lobach (@AntLobach) profile picture

👨‍👩‍👧 Father

Joined May 2024
Photo by Anton Lobach (@AntLobach)Photo by Anton Lobach (@AntLobach)Photo by Anton Lobach (@AntLobach)Photo by Anton Lobach (@AntLobach)Photo by Anton Lobach (@AntLobach)Photo by Anton Lobach (@AntLobach)Photo by Anton Lobach (@AntLobach)Photo by Anton Lobach (@AntLobach)Photo by Anton Lobach (@AntLobach)Photo by Anton Lobach (@AntLobach)
Use pointer velocities to predict the click, then preload the route before it lands. ForesightJS + TanStack Router
64
Stuck with Python but you and your AI agent miss Lisp-style REPL-driven development? Dynra is what you've been longing for. Now with Emacs support. github.com/antlobach/Dynra
4
2
12
706
@AutolithAgent bro, you should try when working on Python projects
1
2
132
Early days asked Claude to draft a release web server for Autolith and the mf started writing Go. In a Lisp repo!!!
2
100
It's always like oh, if you just use modern C++ or if you just enable that compiler flag, or if you just use that feature of C++.
Replying to @_davemacdonald
It’s true. I propose we rewrite all that stuff into modern C++
93
Bro's getting smarter and smarter every day!
Bro I haven't seen anything dumber than OpenClaw, and trust me bro, I am the rock
1
4
252
If only there was a good alternative ... @DelaneyGillilan
Is there any good reason to use all this React and JSX nonsense anymore? Why? Seems like totally unnecessary complexity now. I understand if your enterprise is already locked in, but for anyone building any web app or business today-- is there any reason to choose this stuff?
2
10
543
This is such a nice talk!
6
285
Would be interesting to try gepa-ai.github.io/gepa/blog/… for this
We asked GPT-6 Astra to edit Qwen3.5-0.8B weights to improve math. It directly changed only 0.016% of weights, fixing 0.8B's reasoning loops & improving math up to 6x and even instruction following 3x, without gradients or training. We call this Weight-Space Programming: frontier LLMs can now natively read/write neural network weights! 🧵 Blog: rikiyat.github.io/blog/weigh…
Readers added context they thought people might want to know
LLM was given weights explicitly as text in prompt and used code/forward passes (per authors' logs) to ablate Qwen looping direction, a known thinking-mode issue per model card. Primarily fixed repetition in that mode, not general 'natively' improving weights/math. rikiyat.github.io/blog/weight-sp… github.com/RikiyaT/lm_und… huggingface.co/Qwen/Qwen3.5-0… huggingface.co/RikiyaT/Qwen3.…
1
2
249
Anton Lobach retweeted
and ptyZZZ v0.2.0 is out. - The browser side of a terminal is now a Datastar Rocket web component `<pty-term>`, and - Mouse support!! Terminal apps that use the mouse now get clicks, drags and scrolling, and Shift-drag still selects text.
1
2
1
12
1,848
Anton Lobach retweeted
Migrate from PHP to Python? DM me if you want it done by a proffesional
1
1
5
268
I have received an unprecedented amount of contributions to Autolith from its users. This release is literally just going to have me merging and fixing up stuff, very blessed
2
144
Emacs
I think it’s time to rethink how we build software. ​ We need one IDE that supports every language, is as fast as @zeddotdev, as smart and polished as @jetbrains, and has agentic workflows like @getsome_air (but with a few more features). ​ I’m the best product manager ever.
1
2
21
590
Yeah, it's Clojure and CL. Been there since forever.
Using AI as an excuse to move toward lower level languages feels completely backwards to me “We don’t need to understand all this complexity, AI does” is exactly the wrong direction. Human understanding is required for every functioning software system The goal should be high level languages with low level performance, not low level languages obscured and incomprehensible having to be explained by a non-deterministic AI translation layer What we need is even higher level languages optimized for human understanding. That is the modern bottleneck we’re facing right now
2
24
1,331