Fetching the paper…

Nimble: Efficiently Compiling Dynamic Neural Networks for Model Inference · Around