4 ms·
> While PyTorch is obviously the future in the short term Well, that's not obvious for me. The metrics do not support the death of Keras / TF (https://twitter.
by catwell 4y ago
> While PyTorch is obviously the future in the short term
Well, that's not obvious for me. The metrics do not support the death of Keras / TF (https://twitter.com/fchollet/status/1614322127460782087 https://twitter.com/fchollet/status/1614322127460782087)
François Chollet's Deep Learning with Python is arguably the best book on the topic and teaches people using Keras / TF2. The Keras documentation is excellent as well, and in the long run it will matter. Personally I prefer the way the Keras / TF2 APIs work to PyTorch as well.
Also, I am pretty sure a JAX backend is coming to Keras, and there is TF Lite...
I cannot make a prediction as to which stack will dominate in five years.
- adw 4y agoThe relevant metric is the literature - adoption is driven by new models, and most new models in academia are being implemented in PyTorch or JAX. (In particular, TensorFlow 2 is nowhere to be seen.) The new grads joining the workforce are going to continue to use the tools they are comfortable with and new projects are built around new models.
- grepLeigh 4y agoIf you're familiar with the plumbing/porcelain API paradigm, JAX depends on TensorFlow plumbing (XLA) with a more ergonomic porcelain API. You might not see TensorFlow's plumbing much anymore if you're a new grad running experiments in a notebook, but the "porcelain API" is just the tip of the ice berg of modern machine learning. If you do any work on the JAX framework, you're frequently working with both the JAX and TensorFlow code repositories: https://github.com/google/jax/blob/main/WORKSPACE#L17 https://github.com/google/jax/blob/main/WORKSPACE#L17
- zone411 4y agoYes, I have been reading a lot of papers in the last couple of years and the corresponding code, if available, is using PyTorch much more often than TensorFlow.