[2603.02212] GLEAN: Grounded Lightweight Evaluation Anchors for Contamination-Aware Tabular Reasoning

Computer Science > Databases arXiv:2603.02212 (cs) [Submitted on 22 Jan 2026] Title:GLEAN: Grounded Lightweight Evaluation Anchors for Contamination-Aware Tabular Reasoning Authors:Qizhi Wang View a PDF of the paper titled GLEAN: Grounded Lightweight Evaluation Anchors for Contamination-Aware Tabular Reasoning, by Qizhi Wang View PDF HTML (experimental) Abstract:Tabular reasoning benchmarks mix semantic inference, numerical computation, and brittle table formatting, yet evaluations for small models remain vulnerable to contamination, dataset artifacts, and retrieval failures. We propose GLEAN, a lightweight evaluation protocol that integrates contamination-aware probes, weak-supervision governance, retrieval-reasoning diagnostics, and structured error attribution under tight hardware constraints. We evaluate across TabFact, WTQ via Squall, TableBench, RobuT, and SciTab under a 16GB GPU budget. Using Squall gold SQL as an executable anchor (95.2% execution), GLEAN assigns a deterministic error taxonomy (L0-L4 plus L0.5 context miss) and reveals a stable error-mode separation: TAPEX errors skew toward grounding (L3) while TAPAS errors skew toward hallucination/abstention (L2/L0). We validate evidence-row heuristics against SQL-derived rows on simple queries (0.62 precision / 0.71 recall; hybrid recall 0.81) and show that retrieval Recall@K can saturate even when end-to-end EM/F1 remains limited, motivating attribution beyond raw recall. We release a modular framework with au...