GPT-6 Astra edits 0.016% of Qwen3.5-0.8B weights, boosting math up to 6x without training
sebkrier · x · 2026-11-03
rikiyatakehi's team had GPT-6 Astra directly edit Qwen3.5-0.8B weights to improve math — no gradients or training, changing only 0.016% of weights. It fixed the model's reasoning loops and improved math up to 6x and instruction following 3x. They call it "Weight-Space Programming": frontier LLMs can natively read/write neural network weights.
adamkarvonen pushed back: the tasks are simple and narrow, and Astra is very good at math — almost anything works on simple tasks. Breakout can be solved with a simple algorithm, and Astra merely implemented it in weights. Neat and evidence Astra is smart, but unlikely to generalize to complex tasks.
More from Models
- Zyphra's ZHA brings subquadratic attention via hierarchical tree-structured indexer — yacineMTB · 2026-11-03
- GPT-6 edits Qwen weights directly, lifting MATH-500 from 7.9% to 48.1% with no gradients — KyeGomezB · 2026-11-03
- LiquidAI's LFM3-1.7B tops HF trending with 1k likes and 42.4k downloads in 5 days — maximelabonne · 2026-11-03
- 'Gemini is your friend and alignment is complicated': another eyebrow-raising model reply — vedantmisra · 2026-11-03
- Pedro Domingos: Humans are much more resistant to data than AI is — pmddomingos · 2026-11-03
- OpenAI says search-like queries in ChatGPT are one-fifth of Google's volume — firstadopter · 2026-11-03