why triton server used so many thread in same triton proc?

[soulseen]

Description
running a triton server with the command:

tritonserver --log-info=true --log-warning=true --log-error=true --log-file=engin.log --strict-model-config=false --model-repository=/models

here is the model config:

name: "test"
platform: "tensorflow_savedmodel"
backend: "tensorflow"
max_batch_size: 1024
instance_group [
  {
    count: 0
    kind: KIND_GPU
    gpus:  [0]
  }
 ]
dynamic_batching{
    preferred_batch_size:[ 8 ]
    max_queue_delay_microseconds: 1000

However, when there are only a few requests, the Triton process has more than 200+ threads. Can you please tell me why these many threads exist even though I configured count: 0 in the model config, and how I should limit the number of these threads?
Triton Information
triton version 24.12