Claude修改Gemma 4的8个神经元以调整意大利语时态选择
一名Reddit作者称，他使用Claude Opus 5.5重写了Gemma 4 12B中的8个MLP神经元，在不使用梯度或传统训练的情况下创建了一个根据上下文选择意大利语时态的开关。在作者的两项测试中，该编辑分别留下5%和16%的句子使用原有形式，而系统提示偏好分别为4%和2%。但提示方式破坏了49个控制用例，编辑仅破坏2个；编辑还使200个无关翻译中的182个保持不变，提示方式则为82个。该实验仅涉及一个模型和一种意大利语句式；三次运行中两次成功，第三次因作者测试程序崩溃而失败。
这项自述实验在代码仓库中提供了模型修改、代码、代理日志和基线，是无需梯度或传统训练、针对本地模型单一行为进行定向权重修改的具体案例。其测试范围较窄，第三次运行也未完成，因此结论仅限于作者报告的测试。
Yesterday MIT CSAIL showed GPT-6 Astra editing the weights of a small model by hand. It changed a few neurons of Qwen3.5-0.8B, the model stopped looping on math problems, and its scores jumped. No training at all.
I wanted to try the same thing on something closer to what I actually do with local models, which is translating into Italian. Gemma 4 has a habit I don't love. For sentences like
He had been hiding in the attic for months when the soldiers found him.
it writes «Si era nascosto in soffitta per mesi…». That's correct, but I prefer «Si nascondeva in soffitta da mesi…», which says he was still hiding when they came. The catch is that when the period was already over (he had lived in Paris for ten years, but by 1990 he had moved back), the original tense has to stay. So it's not "always use tense X", it's "use it only when the meaning calls for it".
So I gave Claude Opus 5.5 the same kind of setup MIT used:
- a Python kernel over Gemma's weights;
- 50 forward passes to test its ideas on practice sentences;
- no gradients, no training.