Yes this does dynamic nets, the model is the same as PyTorch, you build a graph for each input(s). I think Cortex is in the Keras-style of model (only layer abs

Yes this does dynamic nets, the model is the same as PyTorch, you build a graph for each input(s). I think Cortex is in the Keras-style of model (only layer abstractions and not "dynamic" in the sense we talk about neural nets) and isn't dynamic, but I could be missing something. The examples seem to all be in that mold: https://github.com/thinktopic/cortex/blob/master/examples/catsdogs-classification/src/catsdogs/training.clj https://github.com/thinktopic/cortex/blob/master/examples/ca...