arXiv cs.AI· Lennart Purucker, Andrej Tschalzev, Nick Erickson, Gioia Blayer, David Holzm\"uller, Alan Arazi, Alexander Pfefferle, Mustafa Tajjar, Ga\"el Varoquaux, Frank Hutter·· 5 小时前AI 评分43
Beyond IID：表格基础模型的泛化能力到底如何
Beyond IID: How General Are Tabular Foundation Models, Really?
AI 导读
一篇 NeurIPS 2026 Datasets and Benchmarks Track 论文推出 BeyondArena，首个支持 IID、时序、分组任务类型及文本、高基数特征的统一表格基准，并配套发布数据集管理框架 DataFoundry。
正文 · 原文
Abstract:Foundation models for predictive machine learning on tabular data have recently gained significant traction in academia and industry. Research communities across disciplines are increasingly evaluating tabular foundation models on diverse datasets and tasks. However, these task- and discipline-specific evaluations remain largely inaccessible to model researchers because benchmark software and evaluation protocols are fragmented. As a result, model researchers rely on standard benchmarks, which are mostly defined for tasks where tabular foundation models already excel. The most challenging scenarios are excluded, limiting meaningful progress in the field by focusing on marginal improvements on IID data rather than on broader, more demanding challenges. To overcome this, we introduce BeyondArena, the first unified holistic benchmark for tabular data that supports diverse task types (IID, temporal, grouped), across sample size and feature dimensionality scales, with diverse feature types (with text, with high cardinality) from a broad range of disciplines. To enable unified benchmarking beyond standard benchmarks, we introduce DataFoundry, a Python framework and metadata schema for curating tabular datasets for predictive machine learning. Our results across 11 models and 142 curated datasets show that current open-source tabular foundation models are competitive or excel on datasets up to 100k rows, while traditional tree-based and deep learning models still dominate on grouped or large data. BeyondArena guides model research for the most demanding challenges in tabular data, enabling progress towards truly foundational tabular models.
| Comments: | Accepted at NeurIPS 2026 Datasets and Benchmarks Track. v2: NeurIPS camera-ready version |
| Subjects: | Machine Learning (cs.LG); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2606.30410 [cs.LG] |
| (or arXiv:2606.30410v2 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2606.30410 arXiv-issued DOI via DataCite |
Submission history
From: Lennart Purucker [view email]
[v1]
Mon, 29 Jun 2026 14:55:52 UTC (4,246 KB)
[v2]
Fri, 30 Oct 2026 13:46:22 UTC (4,153 KB)
来源：arXiv cs.AI · arxiv.org