Alf Viktor Williamsen (@alfviktr) header
Alf Viktor Williamsen (@alfviktr) profile picture

https://nitter.cf/t.co/TPhNX2FZ6X @01dotinc

Mars
Joined March 2021
Photo by Alf Viktor Williamsen (@alfviktr)Photo by Alf Viktor Williamsen (@alfviktr)Photo by Alf Viktor Williamsen (@alfviktr)Photo by Alf Viktor Williamsen (@alfviktr)Photo by Alf Viktor Williamsen (@alfviktr)Photo by Alf Viktor Williamsen (@alfviktr)Photo by Alf Viktor Williamsen (@alfviktr)Photo by Alf Viktor Williamsen (@alfviktr)Photo by Alf Viktor Williamsen (@alfviktr)Photo by Alf Viktor Williamsen (@alfviktr)
Alf Viktor Williamsen retweeted
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.…
2,367
7,260
5,751
55,935
10,051,348
Alf Viktor Williamsen retweeted
The ROI on keeping your mouth shut is extremely high.
200
10,026
350
48,819
1,096,229
«Cooperation…is its own evolutionary force that contributes to an organism’s immediate survival but also creates the possibility for adaptive responses to future challenges.» - Rafe Sagarin
31
Alf Viktor Williamsen retweeted
Turns out the future does have flying cars
47,679
99,259
32,819
830,619
80,306,705
Alf Viktor Williamsen retweeted
Excited to announce that all super intelligence organizations have now jointly agreed to the ultimate in AI/SI safety: Moving all testing to Delta Airlines flights, where accessing the Internet is utterly impossible!
4,801
7,026
920
115,901
8,364,132
Codex resets hits like a drug.
1
37
Reference data tools let an AI agent query the information a task needs at runtime, faster and cheaper. In our study, this took 48% fewer model requests than general SQL. While remaining accurate and reliable. Read the full post here: 01.inc/research/reference-da…
33
The God we humans really abide is the Sun.
53
Alf Viktor Williamsen retweeted
We built high-throughput materials labs in Menlo Park to create a loop between experiments and models. The labs generate fresh data, the models learn from it, and then help us decide what to try next. Using only 1,300 H200s, plus months of our experimental data, we mid-trained and RL’d an open-source model to surpass GPT-6 Astra on our analysis benchmark. We call it Neon. This is real footage from our lab. We’re focusing first on hard problems in materials science, including superconductors, magnets, and semiconductor materials. Read our blog posts below.
274
528
289
5,096
1,674,104
Alf Viktor Williamsen retweeted
"Everything that needs to be said has already been said. But since no one was listening, everything must be said again." - André Gide
8
45
5
254
13,293
Alf Viktor Williamsen retweeted
Commit yourself to these seven virtues: Honor, strength, courage, duty, justice, beauty, and truth.
539
1,606
149
14,943
839,722
Alf Viktor Williamsen retweeted
People killed by AI: zero People killed by governments: millions Let's give control of AI to the governments and call it safety
554
1,964
211
14,710
347,263
«We shall be unable to turn natural advantage to account unless we make use of local guides.» Sun Tzu
28
Alf Viktor Williamsen retweeted
We’re introducing Kylace, for teams whose AI systems cost too much, fall short or need to scale. We bring specialist expertise and take on the practical work of testing improvements against your requirements. This builds on the research and production evaluations we’ve been publishing. Kylace makes that work available to your team. kylace.com/
1
1
25
The cheapest model is not the cheapest model In production, the useful question is not what a model costs per token, but what it costs to do the work.
1
41
If an evaluation can be programmatically verified, you can scale the number of samples. With enough samples, you can make a much higher-confidence model decision. And once you can make that decision with confidence, you can capture the efficiency gains.
1
20
In our case, that meant cutting inference cost by 86% while preserving reliability. Read how on: blog.01.inc/p/the-cheapest-m…
8
Alf Viktor Williamsen retweeted
Deepseek just dropped v4.1 flash, fully open weights it beats gpt 5.6 sol, opus 5 and every chinese model on coding and cybersecurity at ~86x cheaper cost per million tokens running at 420-507 tok/s = faster than gemini 3.8 flash "smallest model in our new architecture family" btw
138
459
70
7,614
343,897