Run TensorFlow model in Torchserve

Hi, I read the Torchserve documentation but didn’t find my answer.
Can we run any type of models and frameworks like TensorFlow and etc. in Torchserve?
or I should convert to ONNX format and write the Inference function?