The 'Astra can read and write weights to improve Qwen' post is misleading

Title: The ‘Astra can read and write weights to improve Qwen’ post is misleading

Stage magicians often do the same trick multiple ways. One iteration uses a gimmick prop, the next one shows you the prop isn’t gimmick but that one palms the object, the third time was a forced choice etc. When you watch the performance you assume there’s only one technique, so each time they do it they seem to debunk your hypothesis about how it must be working.

舞台魔术师经常用多种方式表演同一个戏法。一种迭代使用机关道具，下一次向你展示道具并非机关而是通过手法藏匿物体，第三次则是强制选择等等。当你观看表演时，你会假设只有一种技巧，因此他们每次表演时，似乎都在反驳你关于其运作原理的假设。

This is pretty much how the experiments in the ‘Astra can read and write weights to improve Qwen’ post work. The post’s abstract says: Just by looking at the raw weights, GPT-6 Astra can tell what a network was trained to do. And without any environment feedback, it has no trouble writing, from scratch, small game-playing networks that outplay networks trained with PPO for 100 million steps.

这与“Astra 可以读取和写入权重以改进 Qwen”一文中的实验运作方式非常相似。该文章的摘要称：仅通过查看原始权重，GPT-6 Astra 就能判断网络被训练用于做什么。而且在没有任何环境反馈的情况下，它能够毫无困难地从零开始编写小型游戏网络，其表现优于经过 PPO 训练 1 亿步的网络。

Likewise, asked to make Qwen3.5-0.8B better at math without gradients, Astra identifies an issue in its weights causing reasoning loops that Qwen’s own model card warns about, and fixes this by natively editing 0.016% of the weights. With the same sampling settings, this weight correction raises MATH-500 from 7.9% to 48.1% and generalizes out of the box to instruction following, raising IFEval by 3x.

同样，当被要求在没有梯度的情况下提高 Qwen3.5-0.8B 的数学能力时，Astra 识别出了其权重中导致推理循环的问题（Qwen 自己的模型卡片中对此有警告），并通过原生编辑 0.016% 的权重修复了该问题。在相同的采样设置下，这种权重校正将 MATH-500 的得分从 7.9% 提高到了 48.1%，并能直接推广到指令遵循任务，使 IFEval 提高了 3 倍。

If you read that, would you assume that the Qwen experiment worked ‘just by looking at the raw weights’? Well it doesn’t. There’s experiments that look at raw weights, and experiments that improve a game-playing network, and an experiment that improves Qwen at maths. But those latter two aren’t just working with raw weights, they write Python programs.

如果你读了这些内容，你会认为 Qwen 实验是“仅通过查看原始权重”就成功的吗？事实并非如此。实验中既有查看原始权重的，也有改进游戏网络的，还有改进 Qwen 数学能力的。但后两者不仅仅是在处理原始权重，它们还编写了 Python 程序。

Here’s a summary of what the post actually does and how the techniques work. ¹ Given one recorded episode of the failure. ² In one try, changing 3.5% of the weights. ³ Weights start at zero, so there’s nothing to read. ⁴ Beats PPO-500M only on Breakout (517 vs. 25); loses to it on Seaquest (53 vs. 89) and Asterix (53 vs. 55). ⁵ Runs up to 16 texts and sees only the attention patterns, not the outputs.

以下是该文章实际操作内容及其技术运作方式的总结。¹ 给定一个失败的记录片段。² 在一次尝试中，改变了 3.5% 的权重。³ 权重从零开始，因此没有内容可读取。⁴ 仅在 Breakout 游戏中击败了 PPO-500M（517 对 25）；在 Seaquest（53 对 89）和 Asterix（53 对 55）上输给了它。⁵ 运行多达 16 个文本，仅查看注意力模式，而不查看输出。

The “just by looking at the raw weights” setting (raw floats in context, no tools, no running the network) only appears in experiments 1, 2 and 4, on networks of 17K–172K weights. Every experiment involving a real language model uses Python, and the Qwen result, which is the headline one, also gets to run the model on up to 50 batches of practice problems.

“仅通过查看原始权重”的设置（上下文中为原始浮点数，无工具，不运行网络）仅出现在实验 1、2 和 4 中，且针对的是 1.7 万至 17.2 万权重的网络。每个涉及真实语言模型的实验都使用了 Python，而作为头条结果的 Qwen 实验，也运行了多达 50 批次的练习题。

What the Qwen result actually shows is that Astra can be prompted to devise and perform tasks that a human ML engineer could do. Human ML engineers write snippets of code that analyse or even modify network weights “directly” all the time. Figuring out how to do this productively is not nothing, and AI models of a year or two ago would have failed at all these tasks.

Qwen 的结果实际上表明，可以通过提示 Astra 来设计和执行人类机器学习工程师能够完成的任务。人类机器学习工程师一直在编写用于分析甚至“直接”修改网络权重的代码片段。弄清楚如何高效地做到这一点并非毫无意义，而一两年前的 AI 模型在所有这些任务上都会失败。

But to me the Qwen result and the other experiments that use Python don’t belong in the same post or paper as the experiments that read the weights directly. A single technique that could achieve both, “Author weights for a small game from scratch” and “Improve the performance of a large model on a sample task” would imply a deep capability. If you test a single technique that way you’ve demonstrated a sort of ‘gestalt’ of the weights (it can write them out from scratch) and also that it understands a large existing network. But that’s not at all true here, because it’s not a single technique.

但在我看来，Qwen 的结果以及其他使用 Python 的实验，不应与直接读取权重的实验放在同一篇文章或论文中。如果一种单一技术能够同时实现“从零开始编写小型游戏的权重”和“提高大型模型在样本任务上的性能”，那将意味着一种深层的能力。如果你用这种方式测试单一技术，你就证明了对权重的一种“完形”（它能从零开始写出权重），同时也证明了它理解一个大型现有网络。但这在这里完全不成立，因为它并非单一技术。