Issue 2026-11-12 · Industry · 研究 · 开源 · 安全
New Tools for Exploring the Souls of LLMs

The piece notes mechanistic interpretability has grown from a handful of researchers in 2021 to hundreds of active contributors, while Anthropic and OpenAI face criticism for spending on models rather than safety research. The author built a tool, TinyGPT Visualizer, to help outsiders inspect models.
Read original ↗