GPT-6 edits 0.016% of Qwen3.5-0.8B weights, boosts math up to 6x without training
xuanalogue · x · 2026-11-03
A research team introduces "Weight-Space Programming": frontier LLMs natively reading and writing neural network weights.
- GPT-6 Astra edited Qwen3.5-0.8B's weights to improve math: changing only 0.016% of weights, with no gradients or training, it fixed the 0.8B model's reasoning loops, boosting math up to 6x and instruction following 3x
- @jaydenteoh argues mech interp and optimization as we know them may end — agents will understand weight numerics better than humans, so neuron-by-neuron interpretability effort deserves rethinking
- Work led by @rikiyatakehi, with a detailed blog post
More from Models
- Leaked DeepSeek V4.1-Flash report: 552B MoE, 1M context, KV cache cut to ~1/40 — teortaxesTex · 2026-11-03
- Observers say Dessa eerily matches Karpathy's "summoning ghosts" analogy; Jack calls it fast — analisereal · 2026-11-03
- Together at TC Disrupt: 'token maxing without cost maxing' via smaller models — togethercompute · 2026-11-03
- Dream-1 zero-shot robot test: 100 public episodes show real skills but poor instruction following — paigeinsf · 2026-11-03
- Speculation: DeepSeek to ship V4.1F or V4.2F before any Pro variant — teortaxesTex · 2026-11-03
- Follow-up: Astra's weight edits only worked on Qwen's visible flaw, not on Gemma 270m — teortaxesTex · 2026-11-03