Prosify: Correcting Formatting Bias in LLMs Collection An end-to-end research project on correcting the systematic over-formatting habit of RLHF-trained language models. • 2 items • Updated 4 days ago
Prosify: Correcting Formatting Bias in LLMs Collection An end-to-end research project on correcting the systematic over-formatting habit of RLHF-trained language models. • 2 items • Updated 4 days ago