一种用于表格型基础模型的JEPA方案

arXiv cs.LG 论文
tabular-foundation-models jepa self-supervision machine-learning datasets

摘要

本文提出一种使用联合嵌入预测架构（JEPA）的表格型基础模型训练方案，以解决潜在项坍塌问题，并在147个真实数据集上将其性能与纯值方法进行比较。

arXiv:2609.25541v1 Announce Type: new Abstract: Tabular foundation models learn to predict cell values in context, whereas world-model self-supervision asks for prediction in representation space (LeCun, 2022; Assran et al., 2023). On a tabular foundation-model prior, the latent term of a joint-embedding predictive architecture (JEPA) collapsed in our earlier runs and took the encoder with it to a constant map. We report a recipe under which the latent term survives to convergence beside the value objective: the value head reads the encoder field rather than the predictor, and the target is an exponential moving average (EMA) difference. To bound its cost against the value-only arm, both arms train until a plateau rule stops them, with no fixed step budget. A fixed horizon had confounded a slowdown with a ceiling, since the value-only arm was still improving well past the usual budget. At convergence, in one run per arm, the JEPA arm trails the value-only arm across 147 real datasets, 32:70 wins to losses on classification (29:63 with one entry per dataset name) and 8:24 on regression, the margin small on classification and wider on regression, and the count leans the same way in each stratum and each benchmark. The JEPA arm (jepa) needs 1.42 times as many steps as the value-only arm (ds), and 1.66 times its wall-clock, to reach its plateau.
查看原文
查看缓存全文

缓存时间: 2026/09/23 09:31 

# A JEPA Recipe for Tabular Foundation Models
Source: [https://arxiv.org/abs/2609.25541](https://arxiv.org/abs/2609.25541)
[View PDF](https://arxiv.org/pdf/2609.25541)

> Abstract:Tabular foundation models learn to predict cell values in context, whereas world\-model self\-supervision asks for prediction in representation space \(LeCun, 2022; Assran et al\., 2023\)\. On a tabular foundation\-model prior, the latent term of a joint\-embedding predictive architecture \(JEPA\) collapsed in our earlier runs and took the encoder with it to a constant map\. We report a recipe under which the latent term survives to convergence beside the value objective: the value head reads the encoder field rather than the predictor, and the target is an exponential moving average \(EMA\) difference\. To bound its cost against the value\-only arm, both arms train until a plateau rule stops them, with no fixed step budget\. A fixed horizon had confounded a slowdown with a ceiling, since the value\-only arm was still improving well past the usual budget\. At convergence, in one run per arm, the JEPA arm trails the value\-only arm across 147 real datasets, 32:70 wins to losses on classification \(29:63 with one entry per dataset name\) and 8:24 on regression, the margin small on classification and wider on regression, and the count leans the same way in each stratum and each benchmark\. The JEPA arm \(jepa\) needs 1\.42 times as many steps as the value\-only arm \(ds\), and 1\.66 times its wall\-clock, to reach its plateau\.

## Submission history

From: Mingyu Jeon \[[view email](https://arxiv.org/show-email/f64247af/2609.25541)\] **\[v1\]**Tue, 22 Sep 2026 01:18:41 UTC \(294 KB\)

相似文章

注解版JEPA

Hacker News Top

联合嵌入预测架构（JEPA）用于自监督学习的逐步注解实现与解释，涵盖I-JEPA、V-JEPA和LeJEPA。