multabench/extended-text-reg-anime-planet

REG_TEXT_ANIME_PLANET

Part of MulTaBench, a benchmark for multimodal tabular learning.

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

• Target: Rating Score
• Rows: 14391
• Original source: https://www.kaggle.com/datasets/hernan4444/animeplanet-recommendation-database-2020

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

Column	Type
Name	string
Alternative Name	string
Number Votes	int
Tags	string
Content Warning	string
Type	categorical
Episodes	int
Finished	bool
Duration	float
StartYear	float
EndYear	float
Season	string
Studios	string
Synopsis	string
Rating Score	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},
}