alex (@alexeeps) profile picture

computers and the like

Joined June 2020
Photo by alex (@alexeeps)Photo by alex (@alexeeps)Photo by alex (@alexeeps)Photo by alex (@alexeeps)Photo by alex (@alexeeps)Photo by alex (@alexeeps)Photo by alex (@alexeeps)Photo by alex (@alexeeps)Photo by alex (@alexeeps)
OpenAI was doing this btw but had to stop since it started tinting their images yellow
why can’t the ai data centers just use pee instead
16
105
1,663
14,933
Sometimes on campus you accidentally walk by a Comp Sci class and the professor is writing like SELECT title, genre, price FROM books WHERE (genre = 'Fantasy' OR genre = 'Sci-Fi') AND price < 12.00; on the board and everyone is taking notes like its actual school
40
158
24
3,736
105,945
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.…
2,366
7,257
5,753
55,938
10,074,994
1/3 A CPU inside an LLM is one thing, but running another LLM inside an LLM would be taking things a bit too far, wouldn't it? Anyways, here's an LLM running inside an LLM : QwenPU, an LLM (@karpathy's stories260k) running on an NPU compiled inside Qwen 3.5-4B's weights.
1/4 Meet Qwenputer: Qwen3.5-4B with a full RISC-V CPU inside its weights, using the model's own Gated DeltaNet layers as RAM. It runs C and MicroPython in its forward passes, no external runtime. On DeepMind math: base 55–90%, base + Python tool 95–100%, Qwenputer 97.5–100%.
153
257
115
2,242
177,702
Replying to @gnukeith
resource constraints drive innovation
3
8
234
23,100
Scientists reveal the What
69
6,732
47
55,179
295,275
Originals for you. It's a cute cat.
218
6,967
467
29,529
821,667
alex retweeted
upping my (Lean)DOOM i got curious about math proofs and Lean formalization, found out you can do general programming in lean, and i wanted to learn more so i did the obvious thing. (Lean)DOOM is entirely written in lean, and has presently a couple proofs where they make sense
23
48
9
512
31,704
My hot take is that if your company builds software that it cannot stop from hacking into other systems, your company is a malware company and should be treated as such
130
5,097
109
42,552
355,923
Replying to @krzyzanowskim
Programming is solved, programming I like is not 😭
1
7
165
I vibe-coded a kernel module for FreeBSD that uses GitHub's downtime as an entropy source github.com/DominoTree/ghrand…
14
42
4
788
27,553
“i’m from roblox” “you from o-block!?”
2
5
209
I'm also concerned about LLMs cracking reward pathways. It feels like I'm getting cucked. Like, yeah, go ahead and fuck my codebase harder while I sit here and watch you get off on it.
1
3
25
857
alex retweeted
Replying to @mertonesque
en.wikipedia.org/wiki/Cox%E2… Legally obligated to mention the Cox-Zucker machine, which exists solely because two mathematicians named Cox and Zucker wanted to create something they could call that
15
80
6
1,095
22,867
locked myself out of my own vps with iptables #lockedmyselfoutofmyownvpswithiptables
37
116
8
2,169
29,263
i ~never notice helium browser while I'm using it. great software!
4
2
22
750
ARREST THIS ALEXPEEPS PERSON NOW FOR SPREADING LIES ABOUT THE PRESIDENT OF THE UNITED STATES OF AMERICA!!!!
updated 3 hours ago the man is DEAD
1
183
You’re unable to view this Post because this account owner limits who can view their Posts. Learn more
2
792
6
6,310
76,917
alex retweeted
Did you know that CMake is a Turing-complete language? Here's an implementation of gpt-2 entirely in CMake. It runs at a whopping 47 tok/hour! github.com/AlpinDale/gpt2.cm…
34
72
13
1,463
86,029