Tag
NVIDIA and Stanford developed a Contrastive Language Model (CLM) that frames decision-making in AI as a retrieval problem, achieving up to 9x lower latency compared to existing systems like Jev by caching action embeddings.