Transform ML models into native code with zero dependencies

Re:Transform ML models into native code with zero dependencies by isolli on Tuesday March 05, 2019 @03:54AM (#19309171)

Does anyone know of a similar idea for neural networks? As far as I can tell, you need the entire framework (which requires a heavy, 1.5 GB docker image) to apply a trained model, even though in theory you only need matrix multiplication and a few activation functions.

Related: https://onnx.ai/

[ Parent | on HN ]

Re:Transform ML models into native code with zero dependencies by pplonski86 on Tuesday March 05, 2019 @07:08AM (#19309985)

I wrote keras2cpp https://github.com/pplonski/keras2cpp

It transforms keras + theano models into pure C++, no additional packages. It is not using GPU.

[ Parent | on HN ]

Re:Transform ML models into native code with zero dependencies by ctchocula on Tuesday March 05, 2019 @04:49AM (#19309407)

A similar idea but for neural networks is by using TVM [1]. It's built on top of Halide and LLVM. I haven't tried it myself, but it seems to support deployment to many backends.

[1] https://docs.tvm.ai/deploy/index.html

[ Parent | on HN ]

Re:Transform ML models into native code with zero dependencies by v_lisivka on Tuesday March 05, 2019 @09:39AM (#19310770)

Try Yolo: https://pjreddie.com/darknet/yolo/ . It's written in C.

[ Parent | on HN ]

Re:Transform ML models into native code with zero dependencies by elephantum on Tuesday March 05, 2019 @06:26AM (#19309808)

There’s a project that compiles ONNX models: https://onnc.ai

[ Parent | on HN ]