What if concepts could be directly located?

[luolearning]

Hi — I saw you're using probes to find where concepts live in activations.

I’ve been experimenting with a small plug-in that makes internal activations directly locatable, so concepts don’t need to be inferred via probing.

Curious if this connects to what you're exploring.

https://github.com/luolearning/luoshu_kit