This is great news out of
@huggingface. Similar to Thomas, I also doubt that alignment will easily be solved by the frontier labs. The more alignment research we can do and share in the open, the better open-source models will be for it. Naturally, it is one of our core agenda items at
@baselabs.