multabench/core-img-reg-mkphoto-bots

REG_IMAGE_MKPHOTO_BOTS

Part of MulTaBench, a benchmark for multimodal tabular learning.

Paper: MulTaBench: Benchmarking Multimodal Tabular Learning with Text and Image · Leaderboard: multabench/leaderboard

• Target: TrustNZ_IDNNZ
• Rows: 13748
• Original source: https://www.kaggle.com/datasets/guardeec/mkphoto2023

Load the benchmark splits with multabench.load_split("REG_IMAGE_MKPHOTO_BOTS", fold=0).

Column	Type
gan	int
dtm	int
person	float
n_faces	int
celebs	categorical
BTT	float
speed	float
NBQ	float
image	image path (inside images-*.zip)
TrustNZ_IDNNZ	float

Citation

@misc{arazi2026multabenchbenchmarkingmultimodaltabular,
  title={MulTaBench: Benchmarking Multimodal Tabular Learning with Text and Image},
  author={Alan Arazi and Eilam Shapira and Shoham Grunblat and Mor Ventura and Elad Hoffer and Gioia Blayer and David Holzmüller and Lennart Purucker and Gaël Varoquaux and Frank Hutter and Roi Reichart},
  year={2026},
  eprint={2605.10616},
  archivePrefix={arXiv},
  primaryClass={cs.LG},
  url={https://arxiv.org/abs/2605.10616},
}