Perplexity Publishes Research on Post-Training Models for Search-Accurate Answers

Perplexity released new research detailing its SFT and RL techniques for making models produce more accurate, search-augmented responses.