Vivien Cabannes

Researcher, META

Joined

April 2021

Names

Vivien Cabannes (Preferred)

Emails

****@gmail.com (Confirmed)

,

****@meta.com (Confirmed)

Personal Links

Career & Education History

Researcher

META (meta.com)

2024 – Present

Postdoc

META (meta.com)

2022 – 2024

PhD student

INRIA (inria.fr)

2019 – 2022

Advisors, Relations & Conflicts

No relations added

Expertise

No areas of expertise listed

Publications

Tool use is provably more scalable than in-weight memory for Large Language Models
Sam Houliston, Ambroise Odonnat, Charles Arnal, Vivien Cabannes
- ICLR 2026 Workshop MemAgents
- Readers: Everyone
Efficient RL Training for LLMs with Experience Replay
Charles Arnal, Vivien Cabannes, Taco Cohen, Julia Kempe, Rémi Munos
- SPOT
- Readers: Everyone
Unveiling Simplicities of Attention: Adaptive Long-Context Head Identification
Charles Arnal, Konstantin Donhauser, Mohammad Pezeshki, Vivien Cabannes, David Lopez-Paz, Kartik Ahuja
- Submitted to ICLR 2026
- Readers: Everyone
Asymmetric REINFORCE for off-Policy Reinforcement Learning: Balancing positive and negative rewards
Charles Arnal, Gaëtan Narozniak, Vivien Cabannes, Yunhao Tang, Julia Kempe, Remi Munos
- NeurIPS 2025 poster
- Readers: Everyone
Unveiling Simplicities of Attention: Adaptive Long-Context Head Identification
Konstantin Donhauser, Charles Arnal, Mohammad Pezeshki, Vivien Cabannes, David Lopez-Paz, Kartik Ahuja
- SLLM
- Readers: Everyone
$\mathbb{X}$-Sample Contrastive Loss: Improving Contrastive Learning with Sample Similarity Graphs
Vlad Sobal, Mark Ibrahim, Randall Balestriero, Vivien Cabannes, Diane Bouchacourt, Pietro Astolfi, Kyunghyun Cho, Yann LeCun
- ICLR 2025 Poster
- Readers: Everyone
A Visual Case Study of the Training Dynamics in Neural Networks
Ambroise Odonnat, Wassim Bouaziz, Vivien Cabannes
- Submitted to ICLR 2025
- Readers: Everyone
$\mathbb{X}$-Sample Contrastive Loss: Improving Contrastive Learning with Sample Similarity Graphs
Vlad Sobal, Mark Ibrahim, Randall Balestriero, Vivien Cabannes, Diane Bouchacourt, Pietro Astolfi, Kyunghyun Cho, Yann LeCun
- NeurIPS 2024 Workshop SSL
- Readers: Everyone
$\mathbb{X}$-Sample Contrastive Loss: Improving Contrastive Learning with Sample Similarity Graphs
Vlad Sobal, Mark Ibrahim, Randall Balestriero, Vivien Cabannes, Diane Bouchacourt, Pietro Astolfi, Kyunghyun Cho, Yann LeCun
- ICML 2024 FM-Wild Workshop Poster
- Readers: Everyone
Iteration Head: A Mechanistic Study of Chain-of-Thought
Vivien Cabannes, Charles Arnal, Wassim Bouaziz, Xingyu Alice Yang, Francois Charton, Julia Kempe
- ICML 2024 MI Workshop Poster
- Readers: Everyone

View all 25 publications

Co-Authors

View all 34 co-authors