ethav1

It works by self-generating responses to red-team prompts and self-generating safe corrections to those then using RLHF with the corrections. It isn’t a major departure from traditional RLHF so it is interesting that it performs so much better in this case.

Is matching prompts to agents a popular research problem? Does anyone know of some related work?

CUDA applications are usually written in C++. Also, I think HFT normally works with processing tasks that can’t be done in parallel.

Datasets can be manually curated to contain primarily original images if this becomes a real issue. For example, classifiers can predict whether an image is generated or not. You could adapt the process used to create laion-aesthetic[0] to remove generated images.

[0]: https://github.com/LAION-AI/laion-datasets/blob/main/laion-a...