multabench/extended-img-reg-dvm-car

REG_IMAGE_DVM_CAR

Part of MulTaBench, a benchmark for multimodal tabular learning.

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

• Target: Price
• Rows: 100000
• Original source: https://www.kaggle.com/datasets/osamasaifs/dvm-car-with-images

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

Column	Type
Maker	categorical
Genmodel	string
Adv_year	int
Adv_month	int
Color	categorical
Reg_year	float
Bodytype	categorical
Runned_Miles	float
Engin_size	categorical
Gearbox	categorical
Fuel_type	categorical
Seat_num	float
Door_num	float
Predicted_viewpoint	int
Quality_check	categorical
Car Picture	image path (inside images-*.zip)
Price	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},
}