Nathan Helm-Burger comments on Alignment Can Reduce Performance on Simple Ethical Questions

[Nathan Helm-Burger]

Good work, thanks for doing this.

For future work, you might consider looking into inference suppliers like Hyperdimensional for DeepSeek models.