"Supercharged Interpolation" is not something that actually exists. Learning in High Dimension Always Amounts to Extrapolation https://arxiv.org/abs/2110.0948

"Supercharged Interpolation" is not something that actually exists. Learning in High Dimension Always Amounts to Extrapolation https://arxiv.org/abs/2110.09485 https://arxiv.org/abs/2110.09485 What you're asking for is not "generalization" but magic and humans would also fail.