Realmbird comments on Can we interpret latent reasoning using current mechanistic interpretability tools?