✳flâneur — a map of the web's best reading

Small Models Can Introspect, Too

vgel.me · 7,794 words · saved by 1 readers

Doing introspection experiments with an OSS model.

Small Models Can Introspect, Too Posted December 12, 2025 Recent work by Anthropic showed that Claude models, primarily Opus 4 and Opus 4.1, are able to introspect--detecting when external concepts have been injected into their activations. But not all of us have Opus at home! By looking at the logits, we show that a 32B open-source model that at first appears unable to introspect actually is subtly introspecting. We then show that better prompting can significantly improve introspection performance, and throw the logit lens and emergent misalignment into the mix, showing that the model can in

Explore this link on the map →

related reading