Skip to main content

Read the full write-up

Notion blog post with details
This project explores reducing API costs while maintaining quality in multi-turn search agents. It uses rLLM to train agents that achieve comparable performance with fewer and shorter LLM calls.