Training with another img-size (1280), changes in config_infer_custom_yolo.txt?

[Alberto1404]

@thatbrguy I have trained a custom model using 1280 as img-size during training. For best and most accurate predictions, as far as I know, the inference must take the same img-size. Reason why I exported to ONNX with that same size. This should also be defined in config_infer_custom_yolo file? Does it take any default value (e.g. 640)? If so, the parameter to define in the txt file, based on this website should be infer-dims, right?
That is, config_infer_custom_yolo.txt should contain as follows?:

...
infer-dims = 3;1280;1280
...

Thanks in advance.