arXiv subjects

Eduardo Calò

Publications and source records attributed to Eduardo Calò.

3 recordsLinked to original sources

arXiv2026

Overview of SHROOM-Visions 2026: A Shared Task on Hallucination Detection in Large Vision-Language Models

In 2026, we held the fourth iteration of the SHROOM Shared Task series: SHROOM-Visions (\textbf{S}hared-task on \textbf{H}allucinations and \textbf{R}elated \textbf{O}bservable \textbf{O}vergeneration \textbf{M}istakes in \textbf{Vision} language model\textbf{s}), which is hosted at the UncertaiNLP Workshop co-located with EMNLP 2026. Following the success of the 2024 and 2025 tasks, this time we aim to tackle hallucinations through a model-agnostic detection task focused on large vision-language models. Building on the recently introduced SHEEP dataset, designed for long-term evaluation across model generations, the task invites participants to detect and classify fine-grained hallucination spans in image-conditioned text generation (VQA, image captioning, etc.). The evaluation uses a five-class taxonomy of hallucinations spanning four languages: Chinese, English, French, and Italian. The shared task generated strong interest in the NLP community worldwide, with 27 teams contributing 600+ system submissions. The best systems achieve average scores of 0.58 in character-level correlation, 0.46 in label-conditioned correlation, and 0.51 in intersection-over-union (IoU) across four languages, outperforming the baselines by 30-40 points.

cs.CL↗

arXiv2026

Can Humans Dream of Electric Sheep? Human-Written Samples for Fine-Grained Vision-and-Language Hallucination Benchmarking

In an age of rapid model turnover, how do we make hallucination evaluation more perennial? We explore whether human-written hallucination samples could take the place of model-generated hallucinations, in order to make benchmarking detection independent of particular models. To this end, we construct a dataset of 1,600 human-written samples, spanning four languages (Chinese, English, French, Italian), and 18,400 samples from five vision-and-language models, all annotated for hallucinations using a fine-grained span-level labeling scheme. We find that human-written samples result in higher agreement and allow greater control of dataset contents, while remaining distributionally similar to samples derived from vision-and-language samples and providing a reasonable portrayal of detection capabilities - suggesting that human data is a viable substitute for model-based hallucination benchmarks.

cs.CV↗

arXiv2023

General Boolean Formula Minimization with QBF Solvers

The minimization of propositional formulae is a classical problem in logic, whose first algorithms date back at least to the 1950s with the works of Quine and Karnaugh. Most previous work in the area has focused on obtaining minimal, or quasi-minimal, formulae in conjunctive normal form (CNF) or disjunctive normal form (DNF), with applications in hardware design. In this paper, we are interested in the problem of obtaining an equivalent formula in any format, also allowing connectives that are not present in the original formula. We are primarily motivated in applying minimization algorithms to generate natural language translations of the original formula, where using shorter equivalents as input may result in better translations. Recently, Buchfuhrer and Umans have proved that the (decisional version of the) problem is $Σ_2^p$-complete. We analyze three possible (practical) approaches to solving the problem. First, using brute force, generating all possible formulae in increasing size and checking if they are equivalent to the original formula by testing all possible variable assignments. Second, generating the Tseitin coding of all the formulae and checking equivalence with the original using a SAT solver. Third, encoding the problem as a Quantified Boolean Formula (QBF), and using a QBF solver. Our results show that the QBF approach largely outperforms the other two.

cs.AI↗