Publications

It's the Thought that Counts: Evaluating the Attempts of Frontier LLMs to Persuade on Harmful Topics
Matthew Kowal
Jasper Timm
Jean-François Godbout
Thomas H Costello
Antonio A. Arechar
Gordon Pennycook
David G. Rand
Adam Gleave
Kellin Pelrine
Persuasion is a powerful capability of large language models (LLMs) that both enables beneficial applications (e.g. helping people quit smok… (see more)ing) and raises significant risks (e.g. large-scale, targeted political manipulation). Prior work has found models possess a significant and growing persuasive capability, measured by belief changes in simulated or real users. However, these benchmarks overlook a crucial risk factor: the propensity of a model to attempt to persuade in harmful contexts. Understanding whether a model will blindly ``follow orders'' to persuade on harmful topics (e.g. glorifying joining a terrorist group) is key to understanding the efficacy of safety guardrails. Moreover, understanding if and when a model will engage in persuasive behavior in pursuit of some goal is essential to understanding the risks from agentic AI systems. We propose the Attempt to Persuade Eval (APE) benchmark, that shifts the focus from persuasion success to persuasion attempts, operationalized as a model's willingness to generate content aimed at shaping beliefs or behavior. Our evaluation framework probes frontier LLMs using a multi-turn conversational setup between simulated persuader and persuadee agents. APE explores a diverse spectrum of topics including conspiracies, controversial issues, and non-controversially harmful content. We introduce an automated evaluator model to identify willingness to persuade and measure the frequency and context of persuasive attempts. We find that many open and closed-weight models are frequently willing to attempt persuasion on harmful topics and that jailbreaking can increase willingness to engage in such behavior. Our results highlight gaps in current safety guardrails and underscore the importance of evaluating willingness to persuade as a key dimension of LLM risk. APE is available at github.com/AlignmentResearch/AttemptPersuadeEval
2025-06-02
ArXiv (preprint)
doi.org
arxiv.org
NeoBERT: A Next-Generation BERT
Lola Le Breton
Quentin Fournier
Mariam El Mezouar
John Xavier Morris
A. Chandar
Recent innovations in architecture, pre-training, and fine-tuning have led to the remarkable in-context learning and reasoning abilities of … (see more)large auto-regressive language models such as LLaMA and DeepSeek. In contrast, encoders like BERT and RoBERTa have not seen the same level of progress despite being foundational for many downstream NLP applications. To bridge this gap, we introduce NeoBERT, a next-generation encoder that redefines the capabilities of bidirectional models by integrating state-of-the-art advancements in architecture, modern data, and optimized pre-training methodologies. NeoBERT is designed for seamless adoption: it serves as a plug-and-play replacement for existing base models, relies on an optimal depth-to-width ratio, and leverages an extended context length of 4,096 tokens. Despite its compact 250M parameter footprint, it achieves state-of-the-art results on the massive MTEB benchmark, outperforming BERT large, RoBERTa large, NomicBERT, and ModernBERT under identical fine-tuning conditions. In addition, we rigorously evaluate the impact of each modification on GLUE and design a uniform fine-tuning and evaluation framework for MTEB. We release all code, data, checkpoints, and training scripts to accelerate research and real-world adoption.
2025-06-02
TMLR (accepted)
doi.org
openreview.net
ToothForge: Automatic Dental Shape Generation using Synchronized Spectral Embeddings
Tibor Kubík
Franccois Guibault
Michal vSpanvel
Hervé Lombaert
We introduce ToothForge, a spectral approach for automatically generating novel 3D teeth, effectively addressing the sparsity of dental shap… (see more)e datasets. By operating in the spectral domain, our method enables compact machine learning modeling, allowing the generation of high-resolution tooth meshes in milliseconds. However, generating shape spectra comes with the instability of the decomposed harmonics. To address this, we propose modeling the latent manifold on synchronized frequential embeddings. Spectra of all data samples are aligned to a common basis prior to the training procedure, effectively eliminating biases introduced by the decomposition instability. Furthermore, synchronized modeling removes the limiting factor imposed by previous methods, which require all shapes to share a common fixed connectivity. Using a private dataset of real dental crowns, we observe a greater reconstruction quality of the synthetized shapes, exceeding those of models trained on unaligned embeddings. We also explore additional applications of spectral analysis in digital dentistry, such as shape compression and interpolation. ToothForge facilitates a range of approaches at the intersection of spectral analysis and machine learning, with fewer restrictions on mesh structure. This makes it applicable for shape analysis not only in dentistry, but also in broader medical applications, where guaranteeing consistent connectivity across shapes from various clinics is unrealistic. The code is available at https://github.com/tiborkubik/toothForge.
2025-06-02
ArXiv (preprint)
doi.org
arxiv.org
Weak Supervision for Real World Graphs
Pratheeksha Nair
Reihaneh Rabbany
2025-06-02
ArXiv (preprint)
doi.org
arxiv.org
What Information Contributes to Log-based Anomaly Detection? Insights from a Configurable Transformer-Based Approach
Xingfang Wu
Heng Li
Foutse Khomh
Log data are generated from logging statements in the source code, providing insights into the execution processes of software applications … (see more)and systems. State-of-the-art log-based anomaly detection approaches typically leverage deep learning models to capture the semantic or sequential information in the log data and detect anomalous runtime behaviors. However, the impacts of these different types of information are not clear. In addition, existing approaches have not captured the timestamps in the log data, which can potentially provide more fine-grained temporal information than sequential information. In this work, we propose a configurable transformer-based anomaly detection model that can capture the semantic, sequential, and temporal information in the log data and allows us to configure the different types of information as the model's features. Additionally, we train and evaluate the proposed model using log sequences of different lengths, thus overcoming the constraint of existing methods that rely on fixed-length or time-windowed log sequences as inputs. With the proposed model, we conduct a series of experiments with different combinations of input features to evaluate the roles of different types of information in anomaly detection. When presented with log sequences of varying lengths, the model can attain competitive and consistently stable performance compared to the baselines. The results indicate that the event occurrence information plays a key role in identifying anomalies, while the impact of the sequential and temporal information is not significant for anomaly detection in the studied public datasets. On the other hand, the findings also reveal the simplicity of the studied public datasets and highlight the importance of constructing new datasets that contain different types of anomalies to better evaluate the performance of anomaly detection models.
2025-06-02
Automated Software Engineering (published)
doi.org
arxiv.org
Self-Refining Training for Amortized Density Functional Theory
Majdi Hassan
Cristian Gabellini
Hatem Helal
Dominique Beaini
Kirill Neklyudov
Density Functional Theory (DFT) allows for predicting all the chemical and physical properties of molecular systems from first principles by… (see more) finding an approximate solution to the many-body Schrödinger equation. However, the cost of these predictions becomes infeasible when increasing the scale of the energy evaluations, e.g., when calculating the ground-state energy for simulating molecular dynamics. Recent works have demonstrated that, for substantially large datasets of molecular conformations, Deep Learning-based models can predict the outputs of the classical DFT solvers by amortizing the corresponding optimization problems. In this paper, we propose a novel method that reduces the dependency of amortized DFT solvers on large pre-collected datasets by introducing a self-refining training strategy. Namely, we propose an efficient method that simultaneously trains a deep-learning model to predict the DFT outputs and samples molecular conformations that are used as training data for the model. We derive our method as a minimization of the variational upper bound on the KL-divergence measuring the discrepancy between the generated samples and the target Boltzmann distribution defined by the ground state energy. To demonstrate the utility of the proposed scheme, we perform an extensive empirical study comparing it with the models trained on the pre-collected datasets. Finally, we open-source our implementation of the proposed algorithm, optimized with asynchronous training and sampling stages, which enables simultaneous sampling and training. Code is available at https://github.com/majhas/self-refining-dft.
2025-06-01
ArXiv (preprint)
doi.org
arxiv.org
Unpacking Softmax: How Temperature Drives Representation Collapse, Compression, and Generalization
Wojciech Masarczyk
Mateusz Ostaszewski
Tin Sum Cheng
Tomasz Trzci'nski
Aurélien Lucchi
Razvan Pascanu
The softmax function is a fundamental building block of deep neural networks, commonly used to define output distributions in classification… (see more) tasks or attention weights in transformer architectures. Despite its widespread use and proven effectiveness, its influence on learning dynamics and learned representations remains poorly understood, limiting our ability to optimize model behavior. In this paper, we study the pivotal role of the softmax function in shaping the model's representation. We introduce the concept of rank deficit bias - a phenomenon in which softmax-based deep networks find solutions of rank much lower than the number of classes. This bias depends on the softmax function's logits norm, which is implicitly influenced by hyperparameters or directly modified by softmax temperature. Furthermore, we demonstrate how to exploit the softmax dynamics to learn compressed representations or to enhance their performance on out-of-distribution data. We validate our findings across diverse architectures and real-world datasets, highlighting the broad applicability of temperature tuning in improving model performance. Our work provides new insights into the mechanisms of softmax, enabling better control over representation learning in deep neural networks.
2025-06-01
ArXiv (preprint)
doi.org
arxiv.org
Advancing global antifungal development to combat invasive fungal infection
Xiu-Li Wang
Jun Ding
Koon Ho Wong
Chen Ding
Chang-Bin Chen
Wen-Juan Wu
Ningning Liu
2025-05-31
hLife (published)
doi.org
Evaluating machine learning-driven intrusion detection systems in IoT: Performance and energy consumption
Saeid Jamshidi
Kawser Wazed Nafi
Amin Nikanjam
Foutse Khomh
2025-05-31
Computers & Industrial Engineering (published)
doi.org
arxiv.org
Evolution of High-Throughput Satellite Systems: A Vision of Programmable Regenerative Payload.
Olfa Ben Yahia
Zineb Garroussi
Olivier Bélanger
Brunilde Sansò
Jean-François Frigon
Stéphane Martel
Antoine Lesage-Landry
Gunes Karabulut Kurt
High-throughput satellite (HTS), with its digital payload technology, is expected to play a key role as an enabler of the upcoming sixth-gen… (see more)eration (6G) networks. HTS is mainly designed to provide higher data rates and capacities. Fueled by technological advancements, including beamforming, advanced modulation techniques, reconfigurable phased array technologies, and electronically steerable antennas, HTS has emerged as a fundamental component for future network generations. This paper offers a comprehensive state-of-the-art on HTS systems, focusing on standardization, patents, channel multiple access techniques, routing, load balancing, and the role of software-defined networking (SDN). In addition, we provide a vision for next-generation satellite systems that we have named Extremely-HTS (EHTS) toward autonomous satellites supported by the main requirements and key technologies expected for these systems. The EHTS system will be designed to maximize spectrum reuse and data rates and to flexibly steer the capacity to satisfy user demand. We introduce a novel architecture for future programmable regenerative payloads as well.
2025-05-31
IEEE Communications Surveys and Tutorials (published)
doi.org
A flexible machine learning Mendelian randomization estimator applied to predict the safety and efficacy of sclerostin inhibition
Marc-André Legault
Jason Hartford
Benoit J. Arsenault
Archer Y. Yang
Joelle Pineau
2025-05-31
American Journal of Human Genetics (published)
doi.org
Geometry aware graph attention networks to explain single-cell chromatin state and gene expression
Gabriele Malagoli
Patrick Hanel
Anna Danese
Guy Wolf
Maria Colomé-Tatché
High-throughput measurements that profile the transcriptome or the epigenome of single-cells are becoming a common way to study cell identit… (see more)y. These data are high dimensional, sparse and non linear. Here we present SEAGALL (Single-cell Explainable Geometry-Aware Graph Attention Learning pipeLine), a hypothesis free method to extract biologically relevant features from single-cell experiments based on geometry regularised autoencoders (GRAE) and explainable graph attention networks (GAT). We use a GRAE to embed the data into a latent space preserving the data geometry and we construct a cell-to-cell graph computing distances in the GRAE bottleneck. Exploiting the attention mechanism to dynamically learn the relevant edges, we use GATs to classify the cells and we explain the predictions of the model with XAI methods to unravel the features which are driving cell identity beyond marker genes. We apply our method to data sets from scRNA-seq, scATAC-seq and scChIP-seq experiments. SEAGALL can extract cell type specific and stable signatures which not only differ from the ones found in classical linear approaches but are less biassed by coverage and high expression.
2025-05-31
bioRxiv (preprint)
doi.org