🌟 All leaderboards allow sorting via single-click on column headers.
Closed-Environment Rank v.s. Feature-Shift Rank
Model | Closed-Environment Rank | Feature-Shift Rank |
LightGBM | 4 | 12 |
XGBoost | 3 | 17 |
CatBoost | 2 | 3 |
TabPFN | 9 | 13 |
TabPFNv2 | 5 | 5 |
DANets | 21 | 21 |
MLP | 13 | 7 |
NODE | 23 | 22 |
ResNet | 10 | 8 |
SwitchTab | 17 | 16 |
TabCaps | 18 | 11 |
TabNet | 22 | 23 |
TANGOS | 11 | 4 |
AutoInt | 20 | 20 |
DCNv2 | 12 | 6 |
FT-Transformer | 16 | 18 |
GrowNet | 24 | 25 |
Saint | 15 | 14 |
SNN | 19 | 19 |
TabTransformer | 25 | 24 |
TabR | 7 | 9 |
ModernNCA | 1 | 15 |
Llama3-8B | 14 | 10 |
TabLLM | 6 | 2 |
UniPredict | 8 | 1 |
Feature-Shift Performance Gap
Model | 20% | 40% | 60% | 80% | 100% | 20% | 40% | 60% | 80% | 100% | 20% | 40% | 60% | 80% | 100% |
LightGBM | -0.065 | -0.192 | -0.249 | -0.31 | -0.353 | -0.047 | -0.144 | -0.274 | -0.398 | -0.552 | 0.237 | 0.599 | 0.793 | 1.159 | 1.405 |
XGBoost | -0.076 | -0.218 | -0.261 | -0.328 | -0.375 | -0.043 | -0.125 | -0.228 | -0.342 | -0.496 | 0.233 | 0.592 | 0.84 | 1.197 | 1.49 |
CatBoost | -0.035 | -0.105 | -0.155 | -0.238 | -0.332 | -0.043 | -0.123 | -0.232 | -0.374 | -0.516 | 0.25 | 0.642 | 0.916 | 1.345 | 1.669 |
TabPFN | -0.048 | -0.118 | -0.165 | -0.235 | -0.309 | -0.02 | -0.069 | -0.132 | -0.228 | -0.388 | \\ | \\ | \\ | \\ | \\ |
TabPFNv2 | -0.043 | -0.127 | -0.217 | -0.296 | -0.369 | -0.058 | -0.153 | -0.277 | -0.421 | -0.592 | -0.016 | -0.044 | -0.056 | -0.084 | -0.089 |
DANets | -0.022 | -0.056 | -0.09 | -0.154 | -0.226 | -0.015 | -0.052 | -0.097 | -0.178 | -0.287 | 0.001 | 0.003 | 0.004 | 0.007 | 0.011 |
MLP | -0.024 | -0.062 | -0.107 | -0.185 | -0.271 | -0.023 | -0.065 | -0.123 | -0.203 | -0.36 | 0.028 | 0.076 | 0.128 | 0.184 | 0.25 |
NODE | -0.009 | -0.031 | -0.061 | -0.109 | -0.179 | -0.002 | -0.023 | -0.045 | -0.084 | -0.143 | 0.001 | 0.003 | 0.005 | 0.007 | 0.009 |
ResNet | -0.031 | -0.085 | -0.15 | -0.233 | -0.315 | -0.034 | -0.09 | -0.171 | -0.279 | -0.488 | 0.054 | 0.133 | 0.208 | 0.293 | 0.38 |
SwitchTab | -0.01 | -0.031 | -0.063 | -0.123 | -0.235 | -0.019 | -0.049 | -0.096 | -0.164 | -0.347 | 0.001 | 0.003 | 0.005 | 0.006 | 0.013 |
TabCaps | -0.02 | -0.051 | -0.09 | -0.157 | -0.263 | -0.012 | -0.044 | -0.084 | -0.13 | -0.232 | \\ | \\ | \\ | \\ | \\ |
TabNet | -0.015 | -0.055 | -0.099 | -0.161 | -0.21 | -0.025 | -0.07 | -0.108 | -0.165 | -0.27 | 0.004 | 0.018 | 0.14 | 0.027 | 0.029 |
TANGOS | -0.026 | -0.069 | -0.121 | -0.203 | -0.286 | -0.03 | -0.082 | -0.15 | -0.236 | -0.423 | 0.001 | 0.003 | 0.005 | 0.006 | 0.013 |
AutoInt | -0.028 | -0.065 | -0.115 | -0.188 | -0.269 | -0.015 | -0.071 | -0.145 | -0.262 | -0.383 | 0.012 | 0.034 | 0.063 | 0.095 | 0.163 |
DCNv2 | -0.026 | -0.07 | -0.12 | -0.201 | -0.285 | -0.025 | -0.067 | -0.135 | -0.216 | -0.362 | 0.039 | 0.102 | 0.18 | 0.244 | 0.317 |
FT-Transformer | -0.028 | -0.077 | -0.138 | -0.232 | -0.305 | -0.017 | -0.067 | -0.145 | -0.272 | -0.464 | 0.007 | 0.025 | 0.049 | 0.078 | 0.112 |
GrowNet | -0.01 | -0.041 | -0.077 | -0.138 | -0.198 | -0.008 | -0.026 | -0.045 | -0.077 | -0.105 | 0.003 | 0.005 | 0.009 | 0.016 | 0.018 |
Saint | -0.016 | -0.042 | -0.072 | -0.142 | -0.247 | -0.031 | -0.095 | -0.192 | -0.32 | -0.44 | 0.017 | 0.051 | 0.087 | 0.131 | 0.167 |
SNN | -0.018 | -0.048 | -0.086 | -0.143 | -0.226 | -0.017 | -0.055 | -0.102 | -0.164 | -0.275 | 0.013 | 0.038 | 0.05 | 0.066 | 0.059 |
TabTransformer | -0.016 | -0.044 | -0.078 | -0.133 | -0.203 | -0.009 | -0.032 | -0.056 | -0.086 | -0.15 | 0.001 | 0.002 | 0.002 | 0.003 | 0.006 |
TabR | -0.062 | -0.158 | -0.214 | -0.263 | -0.318 | -0.046 | -0.126 | -0.221 | -0.355 | -0.525 | 0.022 | 0.064 | 0.119 | 0.244 | 0.392 |
ModernNCA | -0.161 | -0.271 | -0.301 | -0.329 | -0.357 | -0.087 | -0.206 | -0.344 | -0.462 | -0.62 | 0.163 | 0.369 | 0.559 | 0.795 | 1 |
Llama3-8B | -0.045 | -0.118 | -0.176 | -0.244 | -0.351 | 0.056 | -0.101 | -0.217 | -0.291 | -0.429 | -0.233 | 0.444 | 0.595 | 0.359 | 0.669 |
TabLLM | -0.022 | -0.042 | -0.056 | -0.066 | -0.08 | -0.007 | -0.017 | -0.103 | -0.314 | -0.245 | \\ | \\ | \\ | \\ | \\ |
UniPredict | -0.019 | -0.038 | -0.058 | -0.074 | -0.094 | -0.135 | -0.137 | -0.123 | -0.139 | -0.176 | \\ | \\ | \\ | \\ | \\ |
Feature-Shift Performance Rank
Model | 0% | 20% | 40% | 60% | 80% | 100% | 0% | 20% | 40% | 60% | 80% | 100% | 0% | 20% | 40% | 60% | 80% | 100% |
LightGBM | 0.838 | 0.811 | 0.695 | 0.645 | 0.589 | 0.550 | 0.800 | 0.640 | 0.589 | 0.496 | 0.412 | 0.308 | 0.975 | 0.886 | 0.785 | 0.692 | 0.618 | 0.539 |
XGBoost | 0.842 | 0.803 | 0.674 | 0.636 | 0.577 | 0.534 | 0.802 | 0.642 | 0.601 | 0.525 | 0.449 | 0.346 | 0.972 | 0.877 | 0.772 | 0.650 | 0.570 | 0.461 |
CatBoost | 0.869 | 0.860 | 0.797 | 0.751 | 0.675 | 0.588 | 0.837 | 0.690 | 0.647 | 0.564 | 0.459 | 0.355 | 0.992 | 0.920 | 0.836 | 0.710 | 0.648 | 0.550 |
TabPFN | 0.831 | 0.823 | 0.757 | 0.718 | 0.652 | 0.587 | 0.785 | 0.609 | 0.597 | 0.550 | 0.489 | 0.383 | \\ | \\ | \\ | \\ | \\ | \\ |
TabPFNv2 | 0.852 | 0.886 | 0.786 | 0.662 | 0.572 | 0.526 | 0.709 | 0.684 | 0.649 | 0.548 | 0.437 | 0.301 | 0.928 | 0.929 | 0.894 | 0.885 | 0.856 | 0.859 |
DANets | 0.753 | 0.783 | 0.753 | 0.725 | 0.670 | 0.605 | 0.535 | 0.558 | 0.542 | 0.505 | 0.462 | 0.379 | 0.666 | 0.481 | 0.580 | 0.509 | 0.580 | 0.549 |
MLP | 0.805 | 0.814 | 0.777 | 0.740 | 0.670 | 0.595 | 0.723 | 0.623 | 0.607 | 0.563 | 0.514 | 0.409 | 0.797 | 0.651 | 0.694 | 0.606 | 0.632 | 0.565 |
NODE | 0.690 | 0.726 | 0.705 | 0.687 | 0.645 | 0.590 | 0.527 | 0.516 | 0.507 | 0.490 | 0.474 | 0.440 | 0.671 | 0.487 | 0.583 | 0.512 | 0.582 | 0.549 |
ResNet | 0.811 | 0.814 | 0.764 | 0.708 | 0.633 | 0.564 | 0.764 | 0.644 | 0.616 | 0.556 | 0.486 | 0.342 | 0.844 | 0.718 | 0.729 | 0.642 | 0.642 | 0.568 |
SwitchTab | 0.797 | 0.807 | 0.787 | 0.762 | 0.708 | 0.616 | 0.747 | 0.628 | 0.620 | 0.582 | 0.543 | 0.422 | 0.662 | 0.474 | 0.574 | 0.501 | 0.574 | 0.541 |
TabCaps | 0.771 | 0.789 | 0.762 | 0.730 | 0.673 | 0.584 | 0.693 | 0.574 | 0.573 | 0.538 | 0.517 | 0.448 | \\ | \\ | \\ | \\ | \\ | \\ |
TabNet | 0.714 | 0.746 | 0.711 | 0.668 | 0.613 | 0.565 | 0.561 | 0.550 | 0.531 | 0.502 | 0.474 | 0.402 | 0.672 | 0.457 | 0.575 | 0.517 | 0.590 | 0.550 |
TANGOS | 0.813 | 0.819 | 0.779 | 0.734 | 0.659 | 0.587 | 0.723 | 0.635 | 0.615 | 0.565 | 0.511 | 0.378 | 0.844 | 0.728 | 0.744 | 0.646 | 0.643 | 0.570 |
AutoInt | 0.765 | 0.796 | 0.762 | 0.719 | 0.654 | 0.579 | 0.527 | 0.580 | 0.549 | 0.498 | 0.429 | 0.333 | 0.726 | 0.551 | 0.630 | 0.546 | 0.599 | 0.525 |
DCNv2 | 0.807 | 0.809 | 0.768 | 0.727 | 0.654 | 0.583 | 0.751 | 0.635 | 0.620 | 0.568 | 0.518 | 0.418 | 0.815 | 0.686 | 0.708 | 0.615 | 0.627 | 0.562 |
FT-Transformer | 0.785 | 0.799 | 0.755 | 0.703 | 0.621 | 0.557 | 0.703 | 0.620 | 0.603 | 0.543 | 0.464 | 0.331 | 0.723 | 0.538 | 0.626 | 0.541 | 0.602 | 0.550 |
GrowNet | 0.689 | 0.735 | 0.708 | 0.679 | 0.628 | 0.573 | 0.404 | 0.454 | 0.450 | 0.433 | 0.418 | 0.379 | 0.675 | 0.493 | 0.587 | 0.515 | 0.581 | 0.547 |
Saint | 0.821 | 0.837 | 0.812 | 0.787 | 0.722 | 0.630 | 0.744 | 0.646 | 0.619 | 0.546 | 0.459 | 0.376 | 0.736 | 0.543 | 0.624 | 0.527 | 0.584 | 0.528 |
SNN | 0.728 | 0.735 | 0.711 | 0.679 | 0.635 | 0.568 | 0.696 | 0.599 | 0.587 | 0.552 | 0.517 | 0.447 | 0.700 | 0.528 | 0.598 | 0.524 | 0.580 | 0.550 |
TabTransformer | 0.643 | 0.718 | 0.690 | 0.665 | 0.617 | 0.551 | 0.406 | 0.532 | 0.512 | 0.492 | 0.478 | 0.419 | 0.662 | 0.484 | 0.578 | 0.511 | 0.576 | 0.543 |
TabR | 0.840 | 0.802 | 0.716 | 0.664 | 0.620 | 0.572 | 0.761 | 0.662 | 0.618 | 0.545 | 0.454 | 0.332 | 0.843 | 0.730 | 0.768 | 0.684 | 0.663 | 0.555 |
ModernNCA | 0.869 | 0.730 | 0.630 | 0.600 | 0.577 | 0.553 | 0.906 | 0.723 | 0.632 | 0.523 | 0.426 | 0.310 | 0.940 | 0.846 | 0.800 | 0.693 | 0.630 | 0.529 |
Llama3-8B | 0.838 | 0.828 | 0.761 | 0.710 | 0.648 | 0.556 | 0.749 | 0.642 | 0.563 | 0.489 | 0.438 | 0.359 | 0.727 | 0.909 | 0.702 | 0.686 | 0.609 | 0.410 |
TabLLM | 0.813 | 0.816 | 0.803 | 0.786 | 0.782 | 0.768 | 0.832 | 0.696 | 0.696 | 0.652 | 0.517 | 0.571 | \\ | \\ | \\ | \\ | \\ | \\ |
UniPredict | 0.790 | 0.831 | 0.809 | 0.798 | 0.779 | 0.763 | 0.829 | 0.702 | 0.700 | 0.708 | 0.701 | 0.674 | \\ | \\ | \\ | \\ | \\ | \\ |