Delta-Audit: Explaining What Changes When Models Change

Model updates (new hyperparameters, kernels, depths, solvers, or data) change performance, but the \emph{reason} often remains opaque. We introduce \textbf{Delta-Attribution} (\mbox{$\Delta$-Attribution}), a model-agnostic framework that explains \emph{what changed} between versions $A$ and $B$ by differencing per-feature attributions: $\Delta\phi(x)=\phi_B(x)-\phi_A(x)$. We evaluate $\Delta\phi$ with a \emph{$\Delta$-Attribution Quality Suite} covering magnitude/sparsity (L1, Top-$k$, entropy), agreement/shift (rank-overlap@10, Jensen--Shannon divergence), behavioural alignment (Delta Conservation Error, DCE; Behaviour--Attribution Coupling, BAC; CO$\Delta$F), and robustness (noise, baseline sensitivity, grouped occlusion). Instantiated via fast occlusion/clamping in standardized space with a class-anchored margin and baseline averaging, we audit 45 settings: five classical families (Logistic Regression, SVC, Random Forests, Gradient Boosting, $k$NN), three datasets (Breast Cancer, Wine, Digits), and three A/B pairs per family. \textbf{Findings.} Inductive-bias changes yield large, behaviour-aligned deltas (e.g., SVC poly$\!\rightarrow$rbf on Breast Cancer: BAC$\approx$0.998, DCE$\approx$6.6; Random Forest feature-rule swap on Digits: BAC$\approx$0.997, DCE$\approx$7.5), while ``cosmetic'' tweaks (SVC \texttt{gamma=scale} vs.\ \texttt{auto}, $k$NN search) show rank-overlap@10$=1.0$ and DCE$\approx$0. The largest redistribution appears for deeper GB on Breast Cancer (JSD$\approx$0.357). $\Delta$-Attribution offers a lightweight update audit that complements accuracy by distinguishing benign changes from behaviourally meaningful or risky reliance shifts.

MEENA (PersianMMMU): Multimodal-Multilingual Educational Exams for N-level Assessment

Recent advancements in large vision-language models (VLMs) have primarily focused on English, with limited attention given to other languages. To address this gap, we introduce MEENA (also known as PersianMMMU), the first dataset designed to evaluate Persian VLMs across scientific, reasoning, and human-level understanding tasks. Our dataset comprises approximately 7,500 Persian and 3,000 English questions, covering a wide range of topics such as reasoning, mathematics, physics, diagrams, charts, and Persian art and literature. Key features of MEENA include: (1) diverse subject coverage spanning various educational levels, from primary to upper secondary school, (2) rich metadata, including difficulty levels and descriptive answers, (3) original Persian data that preserves cultural nuances, (4) a bilingual structure to assess cross-linguistic performance, and (5) a series of diverse experiments assessing various capabilities, including overall performance, the model's ability to attend to images, and its tendency to generate hallucinations. We hope this benchmark contributes to enhancing VLM capabilities beyond English.

Context Awareness Gate For Retrieval Augmented Generation

Retrieval Augmented Generation (RAG) has emerged as a widely adopted approach to mitigate the limitations of large language models (LLMs) in answering domain-specific questions. Previous research has predominantly focused on improving the accuracy and quality of retrieved data chunks to enhance the overall performance of the generation pipeline. However, despite ongoing advancements, the critical issue of retrieving irrelevant information -- which can impair the ability of the model to utilize its internal knowledge effectively -- has received minimal attention. In this work, we investigate the impact of retrieving irrelevant information in open-domain question answering, highlighting its significant detrimental effect on the quality of LLM outputs. To address this challenge, we propose the Context Awareness Gate (CAG) architecture, a novel mechanism that dynamically adjusts the LLMs' input prompt based on whether the user query necessitates external context retrieval. Additionally, we introduce the Vector Candidates method, a core mathematical component of CAG that is statistical, LLM-independent, and highly scalable. We further examine the distributions of relationships between contexts and questions, presenting a statistical analysis of these distributions. This analysis can be leveraged to enhance the context retrieval process in Retrieval Augmented Generation (RAG) systems.

Hidden in Plain Sight: Evaluating Abstract Shape Recognition in Vision-Language Models

Despite the importance of shape perception in human vision, early neural image classifiers relied less on shape information for object recognition than other (often spurious) features. While recent research suggests that current large Vision-Language Models (VLMs) exhibit more reliance on shape, we find them to still be seriously limited in this regard. To quantify such limitations, we introduce IllusionBench, a dataset that challenges current cutting-edge VLMs to decipher shape information when the shape is represented by an arrangement of visual elements in a scene. Our extensive evaluations reveal that, while these shapes are easily detectable by human annotators, current VLMs struggle to recognize them, indicating important avenues for future work in developing more robust visual perception systems. The full dataset and codebase are available at: \url{https://arshiahemmat.github.io/illusionbench/}