Hello, I'm trying the adaptive batching, but it seems not wait the the time I set in max_latency_ms. I set max_latency_ms=2000 for example, and send a request then it just response me immediately, instead of wait for 2 seconds. does anyone know why?

The latency is the "maximum" it will wait. If it has finished processing the request or hit the max batch size, it will be dispatched