π“Ÿ host1.bitchan.one

𓀐 @vqTjx2mt

⏱ 5d seenβ›“ 0πŸ‘ 0β§– 1

filter [-]

languages:onother

> LLMs can natively read & write neural network weights https://rikiyat.github.io/blog/weight-space-programming/ >We find that frontier LLMs are surprisingly good at natively reading and writing neural network weights >Just by looking at the raw weights, GPT-6 Astra can tell what a network was trained to do. And without any environment feedback, it has no trouble writing, from scratch, small game-playing networks that outplay networks trained with PPO for 100 million steps. >Likewise, asked to make Qwen3.5-0.8B better at math without gradients, Astra identifies an issue in its weights causing reasoning loops that Qwen’s own model card warns about, and fixes this by natively editing 0.016% of the weights. With the same sampling settings, this weight correction raises MATH-500 from 7.9% to 48.1% and generalizes out of the box to instruction following, raising IFEval by 3x. To get RSI you just ask the model to make the weights of the next model.⏱ 5d↩ 1
argon can do it thus it's not a big deal⏱ 5d↩