Anthropic just published new research that successfully identified and mapped millions of human-interpretable concepts, called “features”, within the neural networks of Claude. ( www.anthropic.com )
![](https://kbin.chat/media/cache/resolve/entry_thumb/14/87/1487abc112f119c507e534ea3d559429c37a92e9f3b06cfe85bdf42d3c54911e.png)
I often see a lot of people with outdated understanding of modern LLMs....