JAX: Differentiable Computing by Google

Since deep learning took off in the early 2010s, many frameworks were written to facilitate deep learning in both research and production. For the record, let us mention Caffe, Theano, Torch, Lasagne, Tensorflow, Keras or PyTorch. Some of these frameworks disappeared after a while. Others survived and thrive to this day, mainly PyTorch and Tensorflow. Over time, these frameworks evolved into large ecosystems with many different functionalities.