Is this right Sample a random timestep for each image in the train script examples/controlnet/train_controlnet.py?

[Chinese-Coding]

Is this right Sample a random timestep for each image in the train script examples/controlnet/train_controlnet.py?

I am studying the code of train_controlnet.py and I found that the script only samples noise once for each image during training.
According to the definition of diffusion, shouldn't it sample 1000 times or noise_scheduler.config.num_train_timesteps times for each image?
I hope someone can answer this question, I will be very grateful.
The code is here:
https://github.com/huggingface/diffusers/blob/fbff43acc9f52aec18e27806cc258a592f8b53f6/examples/controlnet/train_controlnet.py#L1040C1-L1046C1