It's a foundation model for tabular data that uses in-context learning: feed it your labeled data, and it reads the patterns and returns predictions in a single forward pass. No gradient updates. No per-dataset training. Just data in, predictions out — in seconds.

→ View original post on X — @h2oai,