PyTorch, TensorFlow and JAX. Keras is a popular high-level API that runs on top of them; older frameworks include Caffe and Theano.
Card 1 of 259. Answer: PyTorch, TensorFlow and JAX. Keras is a popular high-level API that runs on top of them; older frameworks include Caffe and Theano.
About · Quizzes · History of AI · To suggest new questions or report an error send me a dm.
Three widely used deep learning frameworks are PyTorch, TensorFlow and JAX. Keras is often named too, though it is a high-level API that runs on top of a backend engine. A framework provides GPU and TPU tensors, automatic differentiation, and building blocks such as layers, losses and optimizers.
PyTorch, created at Facebook (now Meta) and run by the PyTorch Foundation since 2022, is the most common choice in research. Google released TensorFlow in 2015, and it is known for production tools such as TensorFlow Lite. JAX, also from Google, offers NumPy-style code with transformations such as grad, jit and vmap. Keras 3 runs on any of the three. Caffe and Theano mattered early on but are rarely used now.
A common follow-up is static versus dynamic graphs. TensorFlow 1.x had you define a graph and then run it in a session; PyTorch builds the graph as the code runs, which makes debugging easy. TensorFlow 2.0 made eager execution the default and PyTorch 2.0 added torch.compile, so the gap has narrowed.
