arXiv subjects

Fabrice Y Harel-Canada

Publications and source records attributed to Fabrice Y Harel-Canada.

4 recordsLinked to original sources

arXiv2026

Does Reasoning Improve Psychological Depth in Large Language Models? It Depends on Who's Judging

LLM-as-a-Judge evaluators are increasingly used to score open-ended generation, yet a judge's correlation with human ratings on its development set may not guarantee valid measurement when outputs are closely matched and human preferences are subjective. We study this failure mode through psychological depth in short stories. Seven human readers and an LLM-judge ensemble selected on the original scalar Psychological Depth Scale dataset ($ρ= 0.646$) evaluated 60 blinded, prompt-matched story pairs from GPT-5 vs.\ GPT-4o and DeepSeek-R1 vs.\ DeepSeek-V3. Human preferences showed no universal reasoning advantage: GPT-5 was modestly preferred over GPT-4o (60.0--62.9\%), whereas DeepSeek-R1 trailed V3 (42.9\%), and inter-reader agreement was near chance (Krippendorff's $α= 0.070$), with within-reader consistency and recurring weighting patterns suggesting structured heterogeneity rather than random responding. The judge, by contrast, favored reasoning outputs in 89.0\% of dimension-level comparisons and 59 of 60 pairs on aggregate PDS, uniformly across all five evaluator configurations, and its scores were associated with surface features such as sentence length and lexical diversity. These results suggest that development-set performance is insufficient evidence for deployment validity on a shifted distribution, and that point-estimate judges can obscure the heterogeneity in subjective human evaluation.

cs.LG↗

arXiv2026

NormViz: A Benchmark and Framework for Grounding Multimodal Reasoning in Global Cultures

AI systems are used worldwide, but they struggle to serve the needs of culturally diverse populations. Prior work on cultural understanding evaluates AI systems on text-only settings or on visual artifact recognition (e.g. foods, clothing). The ability to reason about visually observable behaviors through local social norms, which we call visual norm understanding, remains unexamined. We introduce NormViz-Bench, a high quality, human-validated benchmark of 3,268 contrastive image pairs (6,536 images) spanning 16 countries. Each pair varies only in the culturally relevant behavior (e.g., objects, attributes, spatial relations, and actions) that alters how each image is interpreted. Each image is labeled as conforming to, violating, or irrelevant to local social norms, and pair-level evaluation requires both images to be correctly classified, thereby preventing reliance on superficial visual shortcuts. Even the strongest VLMs, Gemini 3.0 Flash and Qwen2.5 VL 7B, succeed on only 26.6% and 21.6% of pairs, struggling most with identifying violating and culturally benign visual behaviors. Towards bridging this, we introduce NormViz-Train, a training dataset of 64k images paired with explanations. Though absolute performance remains low (<30%), finetuning on NormViz-Train improves pair accuracy relatively by up to 125% and 36% Qwen3-VL 4B and 8B respectively, showing a path forward to teach models to connect visual perception to cultural significance. Together, NormViz-Bench and NormViz-Train establish visual norm understanding as a challenging and consequential frontier for multimodal AI.

cs.AI↗

arXiv2025

VaPR -- Vision-language Preference alignment for Reasoning

Preference finetuning methods like Direct Preference Optimization (DPO) with AI-generated feedback have shown promise in aligning Large Vision-Language Models (LVLMs) with human preferences. However, existing techniques overlook the prevalence of noise in synthetic preference annotations in the form of stylistic and length biases. To this end, we introduce a hard-negative response generation framework based on LLM-guided response editing, that produces rejected responses with targeted errors, maintaining stylistic and length similarity to the accepted ones. Using this framework, we develop the VaPR dataset, comprising 30K high-quality samples, to finetune three LVLM families: LLaVA-V1.5, Qwen2VL & Qwen2.5VL (2B-13B sizes). Our VaPR models deliver significant performance improvements across ten benchmarks, achieving average gains of 6.5% (LLaVA), 4.0% (Qwen2VL), and 1.5% (Qwen2.5VL), with notable improvements on reasoning tasks. A scaling analysis shows that performance consistently improves with data size, with LLaVA models benefiting even at smaller scales. Moreover, VaPR reduces the tendency to answer "Yes" in binary questions - addressing a common failure mode in LVLMs like LLaVA. Lastly, we show that the framework generalizes to open-source LLMs as editors, with models trained on VaPR-OS achieving ~99% of the performance of models trained on \name, which is synthesized using GPT-4o. Our data, models, and code can be found on the project page https://vap-r.github.io

cs.AI↗

arXiv2025

Sandcastles in the Storm: Revisiting the (Im)possibility of Strong Watermarking

Watermarking AI-generated text is critical for combating misuse. Yet recent theoretical work argues that any watermark can be erased via random walk attacks that perturb text while preserving quality. However, such attacks rely on two key assumptions: (1) rapid mixing (watermarks dissolve quickly under perturbations) and (2) reliable quality preservation (automated quality oracles perfectly guide edits). Through large-scale experiments and human-validated assessments, we find mixing is slow: 100% of perturbed texts retain traces of their origin after hundreds of edits, defying rapid mixing. Oracles falter, as state-of-the-art quality detectors misjudge edits (77% accuracy), compounding errors during attacks. Ultimately, attacks underperform: automated walks remove watermarks just 26% of the time -- dropping to 10% under human quality review. These findings challenge the inevitability of watermark removal. Instead, practical barriers -- slow mixing and imperfect quality control -- reveal watermarking to be far more robust than theoretical models suggest. The gap between idealized attacks and real-world feasibility underscores the need for stronger watermarking methods and more realistic attack models.

cs.CR↗