Hi Everyone,
I am working on an implementation of Ludwig but based on Pytorch (any feedback is welcome):
The goal is to facilitate Pytorch for those that know little about ML on the one hand, and on the other to make experiments easy for those with more advanced knowledge of Pytorch and ML.
Although it is inspired on Ludwig, it’s not only that the soul is on Pytorch, also going for a minimalist approach and more flexible in the sense that anything can be replaced and customized.
You can see it here:
It seems to be functional ATM, and out of the box can model input and output with any of the following data types: (NUMERIC, CATEGORICAL, DATETIME, IMAGE, TEXT, TIME_SERIES)
Currently working on:
- Putting documentation together
- Examples
- A simple abstraction of Botorch/Ax such that optimal hyperparameters can be found in a similar frictionless fashion.
Looking forward to hearing your thoughts,
Best regards,
Jorge