Hallucinations in large language models (LLMs) present a growing challenge across real-world applications, from healthcare to law, where factual reliability is essential. Despite advances in alignment and instruction tuning, LLMs can still generate outputs that are fluent yet fundamentally untrue. Understanding the cognitive dynamics that underlie these hallucinations remains an open problem. In this study, we propose a prompt-based framework to systematically trigger and quantify hallucination: a Hallucination-Inducing Prompt (HIP), which synthetically fuses semantically distant concepts (e.g., periodic table of elements and tarot divination) in a misleading way, and a Hallucination Quantifying Prompt (HQP), which scores the plausibility, confidence, and coherence of the output. Controlled experiments across multiple LLMs revealed that HIPs consistently produced less coherent and more hallucinated responses than their null-fusion controls. These effects varied across models, with reasoning-oriented LLMs showing distinct profiles from general-purpose ones. Our framework provides a reproducible testbed for studying hallucination vulnerability, and opens the door to developing safer, more introspective LLMs that can detect and self-regulate the onset of conceptual instability.

We propose TAROT, a targeted data selection framework grounded in optimal transport theory. Previous targeted data selection methods primarily rely on influence-based greedy heuristics to enhance domain-specific performance. While effective on limited, unimodal data (i.e., data following a single pattern), these methods struggle as target data complexity increases. Specifically, in multimodal distributions, these heuristics fail to account for multiple inherent patterns, leading to suboptimal data selection. This work identifies two primary factors contributing to this limitation: (i) the disproportionate impact of dominant feature components in high-dimensional influence estimation, and (ii) the restrictive linear additive assumptions inherent in greedy selection strategies. To address these challenges, TAROT incorporates whitened feature distance to mitigate dominant feature bias, providing a more reliable measure of data influence. Building on this, TAROT uses whitened feature distance to quantify and minimize the optimal transport distance between the selected data and target domains. Notably, this minimization also facilitates the estimation of optimal selection ratios. We evaluate TAROT across multiple tasks, including semantic segmentation, motion prediction, and instruction tuning. Results consistently show that TAROT outperforms state-of-the-art methods, highlighting its versatility across various deep learning tasks. Code is available at https://github.com/vita-epfl/TAROT.

Large Language Models trained on web-scale text acquire language generation abilities that can solve a wide range of tasks, particularly when task knowledge is refined into the generative prior using in-context examples. However, spurious features learned from noisy data hinder their generalizability. Supervised finetuning can introduce task specificity, but introduce data inefficiency. Prior studies indicate that (i) noisy neural circuitries coexist with generalizable ones within LLMs, and (ii) finetuning typically enhances (or suppresses) existing abilities without introducing newer ones. Building upon these, we propose TaRot, a novel method for task adaptation. TaRot intervenes in the neural circuitries using learnable rotation matrices that are optimized using Bayesian Optimization, on labelled samples in the order of standard few-shot prompting examples. Experiments on multiple classification and generation tasks using LLMs of varying sizes reveal the efficacy of TaRot, improving upon both zero-as well as few-shot performance, with average improvements (across models and tasks) of 23.81% and 11.15%, respectively. Large Language Models (LLMs) acquire the ability to associate different language concepts presented in a sequential context by optimizing the prediction probability of the next token given a context. Despite its apparent simplicity, when scaled across web-sized text corpora, such a learning strategy introduces the ability to solve a wide range of tasks presented in natural language.

Person-job fit is an essential part of online recruitment platforms in serving various downstream applications like Job Search and Candidate Recommendation. Recently, pretrained large language models have further enhanced the effectiveness by leveraging richer textual information in user profiles and job descriptions apart from user behavior features and job metadata. However, the general domain-oriented design struggles to capture the unique structural information within user profiles and job descriptions, leading to a loss of latent semantic correlations. We propose TAROT, a hierarchical multitask co-pretraining framework, to better utilize structural and semantic information for informative text embeddings. TAROT targets semi-structured text in profiles and jobs, and it is co-pretained with multi-grained pretraining tasks to constrain the acquired semantic information at each level. Experiments on a real-world LinkedIn dataset show significant performance improvements, proving its effectiveness in person-job fit tasks.

Telling someone that I gave them an S.T.I. A.I.: It is I, your recent sexual partner. I enjoyed meeting you on Tuesday, July 25, 2023, at Lucky Saloon. Subsequently, an S.T.I. test has yielded a positive result. One in twenty females aged eighteen to twenty-four has an S.T.I., so your exposure is statistically likely, as a sexually active heterosexual male.