People ask me why I don't like Tensorflow, and this post sums up my thoughts: -- it is ridiculous to use Python to then program a strange DSL to assembly an AST of a function. One should just write the function.
new #JuliaLang blog post: On #MachineLearning and #ProgrammingLanguages
Why #MachineLearning and #AI need a full programming language to achieve their full potential instead of reliance on frameworks and libraries.
On Machine Learning and Programming Languages #julialang "Are new ML-tailored languages required, and if so, why? More importantly, what might the ideal ML language of the future look like?"
Interesting read about machine learning and programming languages. ML lets you solve some problems you can't solve via traditional coding. But ML is so much harder to develop and deploy. Imagine a programming language (+ecosystem) designed with ML in mind.
Tooling for AI is at a criticial point, and will have a huge impact on future research
Come see what is going on in #julialang and ML at #sysml. We argue for the need for a programming language approach to ML. Based on
"State-of-the-art models are increasingly programs, with support for programming constructs like loops and recursion, and this brings out many interesting issues in the tools we use to create them..."
"On Machine Learning and Programming Languages" -- an interesting article on the Julia blog! I'd say an alternative to a new lang -- to address shortcomings of e.g., the Python runtime -- would be a flex framework like PyTorch + exporting it via eg ONNX
On Machine Learning and Programming Languages
Another post: "On machine learning and programming languages" - on a language tailored for ML and using Python: #JuliaLang #MachineLearning #Python #TensorFlow