Today on http://keras.io : three MLP models for image classification, demonstrated in less than 300 lines of code in total. MLP-Mixer, FNet, and gMLP.https://keras.io/examples/vision/mlp_image_classification/ …
Show this thread
1. MLP-Mixer is a variation on a depthwise separable convnet 2. FNet is similar to Transformer encoder but replaces attention with a Fourier transform 3. gMLP uses a spatial gating unit
-
-
Great! I will try to reimplement them to learn new things. Thank you for sharing!
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
On point 2, this is getting uncomfortably close to something I've been working on for about a year. The architecture is very similar, and it leverages Fourier transforms, but not in the same way. There's more to it (leveraging Gauge theory, etc), but it outscales transformers.
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.