Large Language Models (LLMs), originally shown to ace various text
comprehension tasks have also remarkably been shown to tackle table
comprehension tasks without specific training. While previous research has
explored LLM capabilities with tabular dataset tasks, our study assesses the
influence of $\textit{in-context learning}$,$ \textit{model scale}$,
$\textit{instruction tuning}$, and $\textit{domain biases}$ on Tabular Question
Answering (TQA). We evaluate the robustness of LLMs on Wikipedia-based
$\textbf{WTQ}$ and financial report-based $\textbf{TAT-QA}$ TQA datasets,
focusing on their ability to robustly interpret tabular data under various
augmentations and perturbations. Our findings indicate that instructions
significantly enhance performance, with recent models like Llama3 exhibiting
greater robustness over earlier versions. However, data contamination and
practical reliability issues persist, especially with WTQ. We highlight the
need for improved methodologies, including structure-aware self-attention
mechanisms and better handling of domain-specific tabular data, to develop more
reliable LLMs for table comprehension.

通过评估 Large Language Models (LLMs) 在基于维基百科和财务报告的 Tabular Question Answering (TQA) 数据集上的表格理解能力，研究发现指令对性能有显著影响，新模型 Llama3 比之前的版本更具鲁棒性，但 WTQ 数据集存在数据污染和实际可靠性问题，需要通过结构感知自注意机制和更好处理特定领域的表格数据来改进 LLMs 的可靠性。