Latent Chain-of-Thought Improves Structured-Data Transformers

• URL: http://arxiv.org/abs/2605.11262v2
• Date: Tue, 19 May 2026 17:22:12 GMT
• Title: Latent Chain-of-Thought Improves Structured-Data Transformers
• Abstract summary: This paper explores latent chain-of-thought and the impact of depth and looping for structured data.<n>We show that latent chain-of-thought is a useful axis for scaling test-time compute for structured data.
• Score: 26.742304109602614
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Chain-of-thought and more broadly test-time compute are known to augment the expressive capabilities of language models and have led to major innovations in reasoning. Motivated by this success, this paper explores latent chain-of-thought as well as the impact of depth and looping for time-series and tabular data. We propose a recurrent scheme in which a structured-data transformer, after an initial forward pass, compresses its query-position hidden states into feedback tokens that are appended to the input and processed again, allowing multiple rounds of latent computation before prediction. We compare CoT models against a same-depth no-CoT baseline, a deeper baseline matched to the CoT model in effective depth, and a looped transformer with weight-tied recurrence but no additional chain-of-thought tokens. Across 36 datasets in time-series forecasting and tabular prediction, latent chain-of-thought improves over the baseline on 7/9 time-series datasets (+12.63\% average gain) and 23/27 tabular datasets (+3.25\% average gain), with CoT models performing best on average in both settings. We also show that the benefit of CoT extends to pretrained foundation models: applying latent CoT to nanoTabPFN, a small open-source tabular foundation model, improves its performance above the much larger TabPFN-v2 on TabArena. Together, these results demonstrate that chain-of-thought is a useful axis for scaling test-time compute for structured data.

Related papers

• Tabby: An Open Pretraining Recipe for Time Series Foundation Models [36.79272249788465]
  Tabby is a long context probabilistic time series foundation model.<n>It adopts an encoder-only patch Transformer architecture and concentrates the contributions on the data and the training procedure.<n>We release training pipeline and model as open source at huawei-noah/trustworthyAI.
  arXiv  Detail & Related papers  (2026-09-12T13:58:39Z)
• TS2TabPFN: Time Series Classification and Extrinsic Regression through Feature Extraction and a Tabular Foundation Model [0.7913770737379631]
  This paper proposes TS2TabPFN, a framework that integrates explicit feature extraction with TabPFN 2.5.<n>We show that TS2TabPFN significantly outperforms state-of-the-art models in TSER tasks with statistical significance.<n>Results suggest that combining foundation models with structured features overcomes single-paradigm limitations.
  arXiv  Detail & Related papers  (2026-08-04T19:32:18Z)
• Tabular Foundation Model for Generative Modelling [34.3599321018728]
  Generative modelling requires robust, holistic representation learning for a given data modality, rather than optimisation for a supervised prediction target alone.<n>Existing tabular foundation generators have not yet consistently matched strong dataset-specific generators in synthetic data quality.<n>We introduce TabFORGE, built on pretrained textbfTabular textbfFOundational textbfRepresentations for textbfGEneration.
  arXiv  Detail & Related papers  (2026-05-10T08:52:28Z)
• Matching Features, Not Tokens: Energy-Based Fine-Tuning of Language Models [102.20309135516186]
  Cross-entropy (CE) training provides dense and scalable supervision for language models.<n>We introduce a feature-matching objective for language-model fine-tuning.<n>To optimize this objective efficiently, we propose energy-based fine-tuning.
  arXiv  Detail & Related papers  (2026-03-12T17:57:50Z)
• End-to-End Compression for Tabular Foundation Models [20.50130399990578]
  We propose TACO, an end-to-end compression model that compresses the training dataset in a latent space.<n>We test our method on the TabArena benchmark, where our proposed method is up to 94x faster in inference time, while consuming up to 97% less memory.
  arXiv  Detail & Related papers  (2026-02-05T13:33:58Z)
• Patch-Level Tokenization with CNN Encoders and Attention for Improved Transformer Time-Series Forecasting [0.0]
  This paper proposes a two-stage forecasting framework that separates local temporal representation learning from global dependency modelling.<n>A convolutional neural network operates on fixed-length temporal patches to extract short-range temporal dynamics and non-linear feature interactions.<n> Token-level self-attention is applied during representation learning to refine these embeddings, after which a Transformer encoder models inter-patch temporal dependencies to generate forecasts.
  arXiv  Detail & Related papers  (2026-01-18T16:16:01Z)
• Reason to Contrast: A Cascaded Multimodal Retrieval Framework [60.99421225506685]
  TTE-v2, a hybrid multimodal retrieval framework, introduces reasoning-driven performance scaling based on additional input token budget rather than model or embedding size.<n>Our approach augments the initial multimodal retrieval with additional reasoning steps for reranking, enabling more expressive query-candidate interactions at test time.<n> Experiments on the MMEB-V2 benchmark demonstrate that TTE-v2-7B achieves a new state-of-the-art accuracy of 75.7%, and that TTE-v2-2B matches or surpasses leading 7B models trained with significantly larger external data.
  arXiv  Detail & Related papers  (2025-12-21T04:52:48Z)
• Mitra: Mixed Synthetic Priors for Enhancing Tabular Foundation Models [85.64873567417396]
  We introduce Mitra, a TFM trained on a curated mixture of synthetic priors selected for their diversity, distinctiveness, and performance on real-world data.<n>Mitra consistently outperforms state-of-the-art TFMs, such as TabPFNv2 and TabICL, across both classification and regression benchmarks.
  arXiv  Detail & Related papers  (2025-10-24T07:15:06Z)
• State-Space Models for Tabular Prior-Data Fitted Networks [1.9815629827604246]
  We investigate the potential of using Hydra, a bidirectional linear-time structured state space model, as an alternative to Transformers in TabPFN.<n>Our experiments show that this approach reduces the order-dependence, achieving predictive performance competitive to the original TabPFN model.
  arXiv  Detail & Related papers  (2025-10-16T11:31:51Z)
• Estimating Time Series Foundation Model Transferability via In-Context Learning [74.65355820906355]
  Time series foundation models (TSFMs) offer strong zero-shot forecasting via large-scale pre-training.<n>Fine-tuning remains critical for boosting performance in domains with limited public data.<n>We introduce TimeTic, a transferability estimation framework that recasts model selection as an in-context-learning problem.
  arXiv  Detail & Related papers  (2025-09-28T07:07:13Z)
• Mind the Gap: Bridging Thought Leap for Improved Chain-of-Thought Tuning [54.65050470296886]
  We propose the CoT Thought Leap Bridge Task, which aims to automatically detect leaps and generate missing intermediate reasoning steps.<n>We demonstrate that models fine-tuned on bridged datasets consistently outperform those trained on original datasets.<n>Our approach effectively enhances distilled data and provides better starting points for reinforcement learning.
  arXiv  Detail & Related papers  (2025-05-20T17:59:31Z)
• A Closer Look at TabPFN v2: Understanding Its Strengths and Extending Its Capabilities [51.08999772842298]
  Tabular Prior-data Fitted Network v2 (TabPFN v2) achieves unprecedented in-context learning performance across diverse downstream datasets.<n>We show that TabPFN v2 can infer attribute relationships even when provided with randomized attribute token inputs.<n>We demonstrate that TabPFN v2's limitations can be addressed through a test-time divide-and-context strategy.
  arXiv  Detail & Related papers  (2025-02-24T17:38:42Z)
• TD3: Tucker Decomposition Based Dataset Distillation Method for Sequential Recommendation [50.23504065567638]
  This paper introduces textbfTD3, a novel textbfDataset textbfDistillation method within a meta-learning framework.<n> TD3 distills a fully expressive emphsynthetic sequence summary from original data.<n>An augmentation technique allows the learner to closely fit the synthetic summary, ensuring an accurate update of it in the emphouter-loop.
  arXiv  Detail & Related papers  (2025-02-05T03:13:25Z)
• Enhancing Foundation Models for Time Series Forecasting via Wavelet-based Tokenization [74.3339999119713]
  We develop a wavelet-based tokenizer that allows models to learn complex representations directly in the space of time-localized frequencies.<n>Our method first scales and decomposes the input time series, then thresholds and quantizes the wavelet coefficients, and finally pre-trains an autoregressive model to forecast coefficients for the forecast horizon.
  arXiv  Detail & Related papers  (2024-12-06T18:22:59Z)
• Loop Neural Networks for Parameter Sharing [1.1049608786515839]
  We introduce a novel Loop Neural Network, which achieves better performance by utilizing longer computational time without increasing the model size. Our approach revisits the input multiple times, refining the prediction by iteratively looping over a subset of the model with residual connections. We demonstrate the effectiveness of this method through experiments comparing versions of GPT-2 with our loop models, showing improved performance in language modeling tasks while maintaining similar parameter counts.
  arXiv  Detail & Related papers  (2024-09-21T17:07:42Z)
• Interpretable Feature Construction for Time Series Extrinsic Regression [0.028675177318965035]
  In some application domains, it occurs that the target variable is numerical and the problem is known as time series extrinsic regression (TSER) We suggest an extension of a Bayesian method for robust and interpretable feature construction and selection in the context of TSER. Our approach exploits a relational way to tackle with TSER: (i), we build various and simple representations of the time series which are stored in a relational data scheme, then, (ii), a propositionalisation technique is applied to build interpretable features from secondary tables to "flatten" the data.
  arXiv  Detail & Related papers  (2021-03-15T08:12:19Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.