MIT CSAIL 发现 GPT-6 Astra 可直接改写模型权重提升性能
MIT 团队让 GPT-6 Astra 直接改 Qwen3.5-0.8B 的权重，只动 0.016% 就把数学分数从 7.9% 拉到 48.1%，完全不用训练，做法挺新奇。
MIT CSAIL 团队让 GPT-6 Astra 直接编辑 Qwen3.5-0.8B 的权重来提升数学能力。整个过程只改动了 0.016% 的权重，无需梯度或训练。该小模型的数学成绩从 7.9% 提升到 48.1%，提高约 6 倍，指令遵循能力也提升 3 倍。团队将这一方法称为 Weight-Space Programming，即前沿 LLM 可以原生读写神经网络权重。
NEW: An MIT CSAIL team found GPT-6 Astra can improve some AI models by rewriting their weights itself, with one edit taking the math score of a small Qwen model to 48.1% from 7.9%. Rikiya Takehi @rikiyatakehi 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… q 🔗 View Quoted Tweet 💬 24 🔄 0 ❤️ 20 👀 4650 📊 17 ⚡
