How GGML format compares to ONNX?

[PawelPerek]

Hi there,
I try to sort machine learning terminology out in my head. Am I right that both ggml and onnx are binary, platform agnostic representation of neural networks? If so, what are use cases for both of them? Is GGML as good in representation other neural architectures like CNN as representing Transformer models?