#end-to-end-latency

0.40s latency 구현한 Encoder-free Early Fusion 기반 Interaction Model

How Thinking Machines built interactivity into the model