MIT提出权重空间编程：前沿大模型可直接读写并生成网络权重

1. 12:10
   AITNT · AI头条

   MIT 探索权重空间编程：让大模型直接修改 Qwen 权重提升数学能力

   研究人员提出权重空间编程方法，让大模型在无需梯度下降和传统训练的情况下直接分析并修改神经网络权重，使 Qwen3.5-0.8B 的 MATH-500 得分从 7.9% 提升至 48.1%。该方法通过大模型诊断前馈网络中的推理循环并仅改动 0.016% 权重，同时展示了从原始权重识别网络用途以及从零编写小游戏网络权重等能力。

2. 10:00
   IT之家 · AI 筛选

   MIT 提出权重空间编程：GPT-6 Astra 仅修改 Qwen3.5-0.8B 约 0.016% 权重使数学成绩提升 6 倍

   麻省理工学院研究人员提出权重空间编程方法，通过让 GPT-6 Astra 在无需梯度下降的情况下直接修改 Qwen3.5-0.8B 约 0.016% 的权重，使其在 MATH-500 测试中的成绩从 7.9% 提升至 48.1%。

3. 09:57
   量子位Selected

   MIT proposes weight-space programming: GPT-6 Astra directly analyzes and edits Qwen3.5 weights to boost performance

   An MIT research team has introduced weight-space programming, allowing the frontier model GPT-6 Astra to directly analyze and modify the weights of other neural networks without gradient-based training. In experiments, Astra identified and edited just 0.016% of the weights in Qwen3.5-0.8B's feedforward networks, resolving reasoning loop issues and boosting its MATH-500 score from 7.9% to 48.1%. In addition, Astra directly generated network weights from scratch for MinAtar benchmark games, achieving a score of 517.4—outperforming a PPO baseline trained for 500 million steps by 20 times.

4. 03:10
   Buzzing HN · 中文精选Selected

   Weight-Space Programming: LLMs Can Directly Read and Write Neural Network Weights to Optimize Qwen3.5-0.8B

   A research team has introduced Weight-Space Programming, discovering that frontier LLMs can natively read and directly modify neural network weights without gradients or environmental feedback.

Current heat 35·Comparable peak 36（Nov 3 13:00）·Comparable change, last 24 hours–

Trends compare only the same parties observed continuously, so the scope can be narrower than current heat counts. Move the pointer or tap the chart to see hourly heat; use the left and right arrow keys on a keyboard.