#tool-using-agents

[ follow ]
#agent-rft #reinforcement-fine-tuning #reward-engineering
Artificial intelligence
fromInfoQ
1 day ago

OpenAI at QCon AI NYC: Fine Tuning the Enterprise

Agent RFT applies reinforcement fine-tuning to tool-using agents to improve multi-step, tool-mediated decision-making via graded rewards and trajectory-level credit assignment.
[ Load more ]