Twitxy Saved

Ivan Skorokhodov

Verified account This account is verified. Learn more

@isskoro

Predicting the next frame @RhodaAI | Previously @Snap @Yandex. I like neural networks and neural networks like me.
Palo Alto, CA skor.sh Joined May 2011
546 Following 3.4K Followers 277 Posts

Latest posts 5

Ivan Skorokhodov Verified account @isskoro
Ok, I was slightly skeptical at first, but then took a bit deeper look and now feel it's completely misleading. The authors simply asked Astra to not do any *gradient descent* training, so guess what? Astra just did gradient-free training instead, wtf.. github.com/RikiyaT/lm_und…
github.com lm_understand-logs/exp8_math/runs/p50c/turns/t105/answer.md at main · RikiyaT/lm_understand-logs Complete logs of the experiments in the blog post 'LLMs can understand and write neural network weights' - RikiyaT/lm_understand-logs
Rikiya Takehi @rikiyatakehi Nov 2
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/weight-sp…
View quoted post on X
13 replies 6 reposts 254 likes 34K views View on X
Ivan Skorokhodov Verified account @isskoro
It's more or less obvious that LLMs can control robots perfectly well if you give them enough time, but "enough time" is exactly the problem. Imagine a human controlling a robot arm via text commands "move left 0.1 cm; rotate the gripper's yaw angle 3 degrees" vs directly tele-operating it with their own hand: the former is fundamentally limited because the high-bandwidth continuous control stream (e.g., humans can generate thousands of "micro-actions" per second) is being squeezed through a low-bandwidth text interface.

So there is no way of speeding up this approach and the stamp "20.7x + thinking pauses removed" would stay there for quite a while until some breakthroughs in LLM acceleration would happen — and that should be something bigger than whatever we've seen in the past ~5 years.

Otherwise, this would result in economically unviable execution in practice and inability to perform low-latency tasks at all (e.g., cooking, humanoid walking, etc.): very soon you will find yourself distilling such an approach into some action-prediction model (which is actually quite interesting) or replacing the text tokens with some hidden representations — and this will be getting closer and closer to a vanilla robotics pipeline.
Jay Chooi @ CoRL 2026 Verified account @chooi_jeq Sep 5
GPT-6 Astra scored 95% on a robot control task, up from Fable 5.1's 40%, with 6.2x fewer output tokens at 2.3x lower cost. 🧵
View quoted post on X
13 replies 13 reposts 147 likes 16K views View on X
Ivan Skorokhodov Verified account @isskoro
Oh gosh, they've just raised $47B 3 months ago. VCs are now subsidizing tech bros token maxxing
Anthropic Verified account @AnthropicAI May 28
We've raised $65 billion in Series H funding at a $965 billion post-money valuation, led by @AltimeterCap, Dragoneer, @Greenoaks, and @sequoia.

This investment will help us advance our research and expand our capacity to meet growing demand for Claude.
View quoted post on X
8 replies 0 reposts 12 likes 1.3K views View on X
Ivan Skorokhodov Verified account @isskoro
If you're going to CVPR this year — we're hosting a small Rhoda AI evening party on Thursday, June 4th. Come say hi if you're into robotics or generative models (or just good food): luma.com/xqrmps2b
luma.com Rhoda AI @ CVPR: Video Models Meet Robotics · Luma Join Rhoda AI for an invite-only evening at The Nickel — one of Denver's best bars, four minutes from the convention center! We'd love to have you there! 🍸…