Linear-programming edit of Qwen3.5-0.8B weights
What happened
On November 11, Hacker News featured Nathan Barry's work on editing Qwen3.5-0.8B model weights with linear programming. The work targets a model reasoning loop and eases it by changing weights, with benchmark results compared against a GPT-6 Astra editing approach. The linear-programming method reportedly needs one-eighth the weight changes of the comparison and lands close to it on benchmarks. The report gives no specific benchmark scores or further test details.
Written by AI from the coverage · updated 14 hours ago
Coverage
Follow the reports to see the story from different sides.
- Hacker News front pageCan a Linear Program Do Weight-Space Programming?
Nathan Barry 用线性规划编辑 Qwen3.5-0.8B 权重，缓解推理循环，以少 8× 的权重改动取得接近 GPT-6 Astra 编辑方案的基准表现。
Heat over time
Heat now 7·Comparable peak 10（Nov 11 22:00）·Comparable change over 24 hours –
The trend only compares accounts observed without gaps, so its range may be smaller than the current heat. Hover or tap the chart for each hour; the left and right arrow keys step through it.