Is it possible to release a binary with the models?

[qwedc001]

As titled.Planning to write an adaptation plugin of your amazing engine to an OCR tools.And I found installing and seperating the project itself from it's dependencies (which is essential to reduce the disk usage of the plugin) difficult.So I was wondering whether this is possible because I noticed the Running in Background in README.

[Mar2ck]

There was a discussion about converting the model to ONNX (#45) but I don't think anyone has figured out how to run it without Pytorch and Transformers yet.