🎉 Introducing... Methods! We are now tracking 730+ building blocks of machine learning: optimizers, activations, attention layers, convolutions and much more! Compare usage over time and explore papers from a new perspective. Browse the catalogue here: paperswithcode.com/methods
The method page gives you an extract from the paper about the method, the architecture, related papers, trends over time, and subcomponents the method relies on. Here is an example for MobileNetV2: paperswithcode.com/method/mobilen…
@paperswithcode This is quite amazing. Now AI system(Hw/framework)designers can take special care to optimize the most important build blocks. This will significantly speed up training.
@paperswithcode That's quite a search space for meta-learning 😀
@paperswithcode @karpathy One of the most influential sites Charting the path of future academic research!!!
@paperswithcode Here something for @codexeditor : if you look into the collection of categories for ML building blocks you can see many have an associated graphical icon . In codex you already have icons or pictures for specific entities. The idea is now :...
@paperswithcode Doing the letter to the three kings, It would be amazing if we could extract just a function or a class with the bulding blocks and not embedded on the code! But it is amazing as it is now! Amazing work that you're doing!
