A task-centric, retrieval-based perspective is offered for how TFMs generalize: it is believed that tabular in-context generalization is largely retrieval-based, and good models are those that learn to identify relevant examples in the provided context and aggregate them well.
Abstract
Tabular Foundation Models (TFMs) increasingly rely on in-context learning, where a model receives labelled examples at inference time and predicts labels for new inputs without updating its weights. Existing TFMs are typically trained on either massive synthetic corpora or very large collections of real datasets. In contrast, we show that surprisingly strong transfer can emerge from self-supervised pre-training on just a single real table. In this setting, we also find that tables tend to be either broadly useful or broadly poor regardless of downstream prediction task, and that the strongest predictor of usefulness is the number of features rather than the number of instances. This leads to a task-centric interpretation of tabular pre-training: the number and the quality of tasks are essential for the pre-training of TFMs. We show that the same task-centric perspective can help corpus design at scale: fine-grained column-level pre-processing consistently improves downstream performance, while no improvements are observed when we filter or deduplicate at the dataset level. Finally, we offer a new perspective for how TFMs generalize: we believe that tabular in-context generalization is largely retrieval-based, and good models are those that learn to identify relevant examples in the provided context and aggregate them well. The mechanics of TFMs have been relatively understudied; our task-centric, retrieval-based perspective offers a new framework to guide future model and corpus design.
The results show that the LLM's capability dissolves with dimension in a way no noise-corrupted classical learner mimics - which explains why LLMs, so capable elsewhere, keep losing to fifty-year-old baselines on tables, while leaving the mechanism of the prediction as an open question.
Tabular foundation models (TFMs) have moved tabular machine learning from per-dataset training towards amortised in-context inference, fitting a small-to-medium table in a single forward pass without a training run. The 2024–2026 release train, the TabPFN and TabICL lines and challengers such as Mitra, LimiX and Orion, has produced a generation whose architectures, capabilities and limits are documented mainly in preprints, while existing surveys treat these models as a subsection of tabular deep learning or of language-model table understanding. This review is, to our knowledge, the first organised around the current generation. From a corpus of 961 screened records and 98 retained studies, it taxonomises the architectures by pretraining regime, maps the capability space across five axes, isolates the language-model-on-tabular strand for prediction, feature engineering and generation, and summarises openness and deployment. A dedicated critical synthesis then reads the reported capabilities against independent evidence: on the studies reviewed here, tree-based and deep models retain the lead across 142 curated datasets that go beyond the standard independent and identically distributed setting; on 112 datasets, the models attain the highest accuracy but weaker conditional coverage than gradient-boosted trees; and robustness under feature shift, fairness and generation quality remain open. Amortised in-context prediction is thus a working paradigm whose independent evidence has yet to match its benchmark claims.
Sergei O. Kurashkin, V. Tynchenko, Alexey S. Borodulin et al.· Machine Learning and Knowled...· 0 citations
Tabular Foundation Models (TFMs) are currently the best approach to tabular prediction problems. They are constructed as transformers that approximate the Bayesian posterior predictive distribution based on a pre-training prior. These univariate predictors can be converted into multivariate ones autoregressively by sampling one target and adding it to the features. However, the faithfulness of the resulting joint has not been investigated. Furthermore, TFMs cannot be evaluated against the posterior itself, at least not on real-world datasets, because the ground-truth distribution is unknown. We therefore propose asking a different question: could a model's predictions result from any joint distribution? To answer this question, we pose two requirements that any such model must satisfy. The first is marginalization consistency, which demands that marginalized conditionals are equal to directly predicted marginals. The second is factorization consistency, which demands that different factorization orders result in equal joint distributions. Every TFM that we evaluate violates both of these requirements for both classification and regression across all datasets.
Christian Klötergens, Vijaya Krishna Yalavarthi, Lars Schmidt-Thieme et al.· 2 citations
Tabular foundation models, driven by in-context learning, have rapidly grown in quality and popularity. However, recent approaches with either cell-based architectures or retrieval have sacrificed efficiency for raw performance, restricting their utility in situations where compute is limited or inference speed is crucial. We adopt an alternate approach, sticking with row-based attention while incorporating long context pre-training to eliminate the need for retrieval. By combining this with architectural improvements and SSL pre-training on a newly-sourced, larger corpus of real data results, we present TabDPT-Turbo, a model that provides comparable default performance to TabDPT v1.1 on TabArena-Lite, CC18, and CTR23, at orders of magnitude faster. In our experiments, TabDPT-Turbo is the fastest model overall among leading foundation models. We have released the new model as TabDPT v1.2 at https://github.com/layer6ai-labs/TabDPT-inference.
Rasa Hosseinzadeh, Alex Labach, Zexin Xue et al.· 2 citations
Learning to rank (LTR) traditionally requires large-scale training data to generalize effectively. In low-data domains where expert annotation is scarce, the performance of LTR methods degrades sharply. Foundation models have alleviated similar data dependencies in other domains via in-context learning, but a foundation model for ranking with tabular features has not been explored yet. We propose prior-data fitted networks (PFNs) as a strong method for ranking in low-data settings. First, we demonstrate that PFNs, which are originally trained for classification, successfully outperform classification baselines on ranking data. Next, we evaluate PFNs as rankers, showing that they surpass state-of-the-art tuned baselines in low-data regimes. We introduce a novel sampling and inference scheme to obtain pairwise predictions from PFNs' native pointwise architecture, analogous to pairwise LTR. To address the limited context window of the transformers underlying PFNs, we propose a dynamic support set selection strategy for queries that scales PFNs beyond random subsampling. Our experimental results show that PFNs are an effective foundation model for ranking that provides significant gains when data is limited.
David Vos, Samarth Bhargav, M. D. Rijke et al.· International Conference on...· 0 citations
Many text classification decisions are viable based on constituent excerpts alone. Taking inspiration from the field of multiple instance learning, we present an algorithm for training a neural network to classify text by selecting such excerpts. We show that our approach is also scalable with demonstrated learning against samples with nearly 1M tokens. We evaluate our methods on 7 datasets with emphasis on long-textual collections that far exceed the encoding limit of our base model. We present state-of-the-art results with this algorithm on 3 datasets: identification of political bias in news outlets, trigger warnings in long stories, and demographic characteristics of authors in tweet collections. Furthermore, the model trained on weakly-labeled collections of text (bags) generalizes to accurately classify constituent, smaller instances. Besides a new state-of-the-art for these problems, this approach is one of the few neural methods to excel in these datasets.
John J. Cadigan, Dayne Freitag, Eric Yeh· 0 citations
Related blog posts
MIT News · Artificial Intelligence· news.mit.eduAug 27, 2026
A new machine-learning framework aims to improve the success rate of computational protein design while moving away from results that reproduce sequences found in nature.
MIT News · Artificial Intelligence· news.mit.eduAug 24, 2026
A new method for surgically removing training examples from a model reveals that as datasets grow, the link between what a model learns and what it produces dissolves.
