Is there an easy way to port any PyTorch model into GGML?

[Connor56]

I want to run a BERT-like model on my CPU using GGML to see if it's faster than the PyTorch version. I can't find any documentation on how to use GGML. What's the easiest way of doing something like this?

[ggerganov]

Try this tutorial and see if it works for your model: ggml-org/llama.cpp#7712