Wassim (Wes) Bouaziz

Publications and projects. You can also find most up-to-date entries for my latest publications on the Google Scholar page.

Cultivating Pluralism In Algorithmic Monoculture: The Community Alignment Dataset teaser

Cultivating Pluralism In Algorithmic Monoculture: The Community Alignment Dataset

{"name" => "Lily Hong Zhang", "url" => "https://lhz1029.github.io/"}{"name" => "Smitha Milli", "url" => "https://smithamilli.com/"}{"name" => "Karen Jusko"}{"name" => "Jonathan SMith"}{"name" => "Brandon Amos", "url" => "https://bamos.github.io/"}{"name" => "Wassim (Wes) Bouaziz", "url" => "https://wesbz.github.io"}{"name" => "Manon Revel", "url" => "https://manrev.github.io/manon/"}{"name" => "Jack Kussman"}{"name" => "Lisa Titus"}{"name" => "Bhaktipriya Radharapu"}{"name" => "Jane Yu", "url" => "https://janeyubear.github.io/"}{"name" => "Vidya Sarma"}{"name" => "Kris Rose"}{"name" => "Maximilian Nickel", "url" => "https://maxn.io/"}
2nd Workshop on Models of Human Feedback for AI Alignment (MoFA) ICML 2025, Jul, 2025
On Monotonicity in AI Alignment teaser

On Monotonicity in AI Alignment

{"name" => "Gilles Bareilles", "url" => "https://gbareilles.fr/"}{"name" => "Julien Fageot"}{"name" => "Lê-Nguyên Hoang"}{"name" => "Peva Blanchard"}{"name" => "Wassim (Wes) Bouaziz", "url" => "https://wesbz.github.io"}{"name" => "Sébastien Rouault", "url" => "https://sebastien.rouau.lt/"}{"name" => "El Mahdi El Mhamdi", "url" => "https://elmahdielmhamdi.com/"}
arXiv preprint arXiv:2506.08998, Jun, 2025
Winter Soldier: Backdooring Language Models at Pre-Training with Indirect Data Poisoning teaser

Winter Soldier: Backdooring Language Models at Pre-Training with Indirect Data Poisoning

{"name" => "Wassim (Wes) Bouaziz", "url" => "https://wesbz.github.io"}{"name" => "Nicolas Usunier"}{"name" => "El Mahdi El Mhamdi", "url" => "https://elmahdielmhamdi.com/"}
ICLR 2025 Workshop on GenAI Watermarking (WMARK), Feb, 2025
Clustering Head: A Visual Case Study of the Training Dynamics in Transformers teaser

Clustering Head: A Visual Case Study of the Training Dynamics in Transformers

{"name" => "Ambroise Odonnat*", "url" => "https://ambroiseodt.github.io/"}{"name" => "Wassim (Wes) Bouaziz*", "url" => "https://wesbz.github.io"}{"name" => "Vivien Cabannes", "url" => "https://viviencabannes.github.io/"}
arXiv preprint arXiv:2410.24050, Feb, 2025
Easing Optimization Paths: a Circuit Perspective teaser

Easing Optimization Paths: a Circuit Perspective

{"name" => "Ambroise Odonnat*", "url" => "https://ambroiseodt.github.io/"}{"name" => "Wassim (Wes) Bouaziz*", "url" => "https://wesbz.github.io"}{"name" => "Vivien Cabannes", "url" => "https://viviencabannes.github.io/"}
Lecture at IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) 2025, Jan, 2025
Targeted Data Poisoning for Black-Box Audio Datasets Ownership Verification teaser

Targeted Data Poisoning for Black-Box Audio Datasets Ownership Verification

{"name" => "Wassim (Wes) Bouaziz", "url" => "https://wesbz.github.io"}{"name" => "El Mahdi El Mhamdi", "url" => "https://elmahdielmhamdi.com/"}{"name" => "Nicolas Usunier"}
IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) 2025, Nov, 2024
Data Taggants: Dataset Ownership Verification via Harmless Targeted Data Poisoning teaser

Data Taggants: Dataset Ownership Verification via Harmless Targeted Data Poisoning

{"name" => "Wassim (Wes) Bouaziz", "url" => "https://wesbz.github.io"}{"name" => "Nicolas Usunier"}{"name" => "El Mahdi El Mhamdi", "url" => "https://elmahdielmhamdi.com/"}
International Conference on Learning Representations (ICLR) 2025, Oct, 2024
Iteration Head: A Mechanistic Study of Chain-of-Thought teaser

Iteration Head: A Mechanistic Study of Chain-of-Thought

{"name" => "Vivien Cabannes", "url" => "https://viviencabannes.github.io/"}{"name" => "Charles Arnal", "url" => "https://charlesarnal.github.io/"}{"name" => "Wassim (Wes) Bouaziz", "url" => "https://wesbz.github.io"}{"name" => "Alice Yang"}{"name" => "Francois Charton", "url" => "https://f-charton.github.io/about/"}{"name" => "Julia Kempe", "url" => "https://cims.nyu.edu/~kempe/"}
Advances in Neural Information Processing Systems (NeurIPS) 2024, Oct, 2024
The Llama 3 Herd of Models teaser

The Llama 3 Herd of Models

{"name" => "Llama 3 Team"}
arXiv preprint arXiv:2407.21783, Jul, 2024
Inverting Gradient Attacks Makes Powerful Data Poisoning teaser

Inverting Gradient Attacks Makes Powerful Data Poisoning

{"name" => "Wassim (Wes) Bouaziz", "url" => "https://wesbz.github.io"}{"name" => "El Mahdi El Mhamdi", "url" => "https://elmahdielmhamdi.com/"}{"name" => "Nicolas Usunier"}
arXiv preprint arXiv:2410.21453, Feb, 2023
On the Parameterized Complexity of Counting Small-Sized Minimum $\boldsymbol{(S, T)}$-Cuts teaser

On the Parameterized Complexity of Counting Small-Sized Minimum $\boldsymbol{(S, T)}$-Cuts

{"name" => "Pierre Bergé", "url" => "https://perso.isima.fr/~piberge/index.html"}{"name" => "Wassim (Wes) Bouaziz", "url" => "https://wesbz.github.io"}{"name" => "Arpad Rimmel", "url" => "https://www.centralesupelec.fr/chercheurs/arpad-rimmel"}{"name" => "Joanna Tomasik", "url" => "https://www.centralesupelec.fr/chercheurs/joanna-tomasik"}
SIAM Journal on Discrete Mathematics, Jan, 2023
$\texttt{pyannote.audio}$: neural building blocks for speaker diarization teaser

$\texttt{pyannote.audio}$: neural building blocks for speaker diarization

{"name" => "Hervé Bredin", "url" => "https://herve.niderb.fr/"}{"name" => "Ruiqing Yin"}{"name" => "Juan Manuel Coria", "url" => "https://juanmc2005.github.io/"}{"name" => "Gregory Gelly"}{"name" => "Pavel Korshunov", "url" => "https://www.idiap.ch/~pkorshunov/"}{"name" => "Marvin Lavechin", "url" => "https://marvinlvn.github.io/"}{"name" => "Diego Fustes"}{"name" => "Hadrien Titeux"}{"name" => "Wassim (Wes) Bouaziz", "url" => "https://wesbz.github.io"}{"name" => "Marie-Philippe Gill"}
IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) 2020, Nov, 2019