两篇新研究论文探讨了表格基础模型的复杂性。一项研究调查了这些模型的推理动态，揭示了显著的深度冗余，并提出了一种更高效的单层架构。另一篇论文比较了表格模型的不同预训练语料库，发现像TabICL这样的合成数据源占据了真实世界数据分布的一个狭窄区域，并且精心策划的数据和网络抓取的数据在很大程度上是可互换的。 AI
影响 这些研究为优化表格模型的效率和理解预训练数据分布的影响提供了见解。
排序理由 两篇arXiv论文提出了关于表格基础模型的新研究发现。
AI 生成摘要 · Google Gemini · 来自 4 个来源。 我们如何撰写摘要 →
arXiv:2605.06510v1 Announce Type: new Abstract: Transformer-based tabular foundation models (TFMs) dominate small to medium tabular predictive benchmark tasks, yet their inference mechanisms remain largely unexplored. We present the first large-scale mechanistic study of layerwis…
arXiv:2605.06343v1 Announce Type: new Abstract: Tabular foundation models are pre-trained on one of three classes of corpus: curated datasets drawn from benchmark repositories, tables harvested at scale from the web, or synthetic tables sampled from a parametric generative prior.…
Transformer-based tabular foundation models (TFMs) dominate small to medium tabular predictive benchmark tasks, yet their inference mechanisms remain largely unexplored. We present the first large-scale mechanistic study of layerwise dynamics in 6 state-of-the-art tabular in-cont…
Tabular foundation models are pre-trained on one of three classes of corpus: curated datasets drawn from benchmark repositories, tables harvested at scale from the web, or synthetic tables sampled from a parametric generative prior. Despite the centrality of pre-training data to …