Great series of posts, it's formed a useful intro to the tabular foundation models for me. I hope to play around with them myself, and will likely use your examples as a starting point.
A strange thing I noticed about TabPFN is that the order of the input columns has a small effect on performance. And hey, someone already seemed to have picked this up: https://arxiv.org/html/2502.06684v1 .
Great series of posts, it's formed a useful intro to the tabular foundation models for me. I hope to play around with them myself, and will likely use your examples as a starting point.
Appreciate this Christoph. What a different paradigm.
A strange thing I noticed about TabPFN is that the order of the input columns has a small effect on performance. And hey, someone already seemed to have picked this up: https://arxiv.org/html/2502.06684v1 .