TableMind++: An Uncertainty-Aware Programmatic Agent for Tool-Augmented Table Reasoning

TL;DR

TableMind++ enhances programmatic table reasoning by integrating uncertainty-aware inference techniques, significantly reducing hallucinations and improving accuracy through memory-guided plan pruning, confidence-based action refinement, and dual-weighted trajectory aggregation.

Contribution

This work introduces an uncertainty-aware inference framework for programmatic agents, combining memory-guided plan pruning, confidence-based action refinement, and trajectory aggregation to improve reasoning accuracy.

Findings

Outperforms previous baselines and proprietary models on diverse benchmarks.

Effectively mitigates hallucinations in table reasoning tasks.

Demonstrates robustness and improved accuracy with uncertainty quantification.

Abstract

Table reasoning requires models to jointly perform semantic understanding and precise numerical operations. Most existing methods rely on a single-turn reasoning paradigm over tables which suffers from context overflow and weak numerical sensitivity. To address these limitations, we previously proposed TableMind as a tuning-based autonomous programmatic agent that simulates human-like interaction within a lightweight large language model (LLM). TableMind internalizes planning, action, and reflection through a two-stage training strategy involving supervised fine-tuning (SFT) on filtered high-quality data and reinforcement learning (RL) via a multi-perspective reward and the Rank-Aware Policy Optimization (RAPO) algorithm. While TableMind establishes a solid foundation for programmatic agents, the inherent stochasticity of LLMs remains a critical challenge that leads to hallucinations. In this paper, we extend this foundation to TableMind++ by introducing a novel uncertainty-aware inference framework to mitigate hallucinations. Specifically, we propose memory-guided plan pruning to retrieve historical trajectories for validating and filtering out logically flawed plans to address epistemic uncertainty. To ensure execution precision, we introduce confidence-based action refinement which monitors token-level probabilities to detect and self-correct syntactic noise for aleatoric uncertainty mitigation. Finally, we employ dual-weighted trajectory aggregation to synthesize a robust consensus from multiple reasoning paths. Extensive experiments on diverse benchmarks demonstrate that TableMind++ consistently outperforms previous baselines and proprietary models to validate the effectiveness of integrating autonomous training with uncertainty quantification. Our code is available.