For machine learning (un)fortunately, lots of the stack runs on Python. Lots of ceremony was done to circumvent GIL (e.g. PyTorch data loader). “Reasonable perf

For machine learning (un)fortunately, lots of the stack runs on Python. Lots of ceremony was done to circumvent GIL (e.g. PyTorch data loader). “Reasonable performance Python” I imagine is actually something in huge demand for lots of ML shops