https://mikotohhh.github.io/Recent content on Luozhong ZhouHugoen-usWed, 22 Jul 2026 00:00:00 +0000https://mikotohhh.github.io/posts/jspace-cot-tradeoff/Wed, 22 Jul 2026 00:00:00 +0000https://mikotohhh.github.io/posts/jspace-cot-tradeoff/I tested whether Anthropic&rsquo;s finding—that written reasoning protects a model from J-space interventions—also holds for Qwen3-4B. I did not reproduce the math result, and follow-up experiments showed why that null result remains inconclusive.https://mikotohhh.github.io/posts/auditable-malware-analysis/Wed, 22 Jul 2026 00:00:00 +0000https://mikotohhh.github.io/posts/auditable-malware-analysis/A real compromise became a case study in supervising capable agents: bounded autonomy, auditable evidence, and claims designed to fail closed.https://mikotohhh.github.io/about/Mon, 01 Jan 0001 00:00:00 +0000https://mikotohhh.github.io/about/<p>I&rsquo;m <strong>Luozhong Zhou</strong>, a machine learning researcher. I write here about LLMs, interpretability, and the craft of running careful experiments.</p> <ul> <li>GitHub: <a href="https://github.com/mikotohhh">mikotohhh</a></li> <li>Email: luozhong [at] sakana.ai</li> </ul>