Interpretability techniques like probes, steering, and J-lens carry over to DiffusionGemma with performance largely retained

we also examined how interpretability techniques carry over to DiffusionGemma, including probes, steering, and J-lens. We find that performance is largely retained.

https://www.alignmentforum.org/posts/QBuJ3suRZxrrxSTtv/does-diffusiongemma-do-latent-reasoning