Razvan Pascanu

Affiliate Member, Mila - Quebec Artificial Intelligence Institute

Research Scientist, Google DeepMind

Joined

September 2016

Names

Razvan Pascanu (Preferred)

Emails

****@google.com (Confirmed)

****@gmail.com (Confirmed)

Personal Links

Career & Education History

Affiliate Member

Mila - Quebec Artificial Intelligence Institute (mila.quebec)

2024 – Present

Research Scientist

Google DeepMind (google.com)

2015 – Present

PhD student

Université de Montréal (umontreal.ca)

2009 – 2014

Advisors, Relations & Conflicts

PhD Advisor

Yoshua Bengio

2009 – 2014

Expertise

Continual Learning/Lifelong learning

2015 – 2024

Graph Nets

2015 – 2024

Deep Reinforcement Learning

2014 – 2024

Optimization

2009 – 2024

Deep Learning

2007 – 2024

RNN

2007 – 2024

Publications

Loss Smoothing for Continual Adaptation
Darshan Patil, Ekaterina Lobacheva, Razvan Pascanu, Sarath Chandar
- CAO Poster
- Readers: Everyone
How Overconfidence in Initial Choices and Underconfidence Under Criticism Modulate Change of Mind in Large Language Models
Dharshan Kumaran, Stephen M. Fleming, Larisa Markeeva, Joseph Heyward, Mrinal Mathur, Andrea Banino, Razvan Pascanu, Simon Osindero, Benedetto De Martino, Petar Veličković, Viorica Patraucean
- Arxiv
- Readers: Everyone
A Survey on Over-smoothing and Over-squashing: Unified Propagation Perspectives on Graph Neural Networks and Transformers
Alvaro Arroyo, Federico Barbero, Hugh Blayney, Michael M. Bronstein, Xiaowen Dong, Pietro Lio, Razvan Pascanu, Pierre Vandergheynst
- Accepted by TMLR
- Readers: Everyone
TRecViT: A Recurrent Video Transformer
Viorica Patraucean, Xu Owen He, Joseph Heyward, Chuhan Zhang, Mehdi S. M. Sajjadi, George-Cristian Muraru, Artem Zholus, Mahdi Karami, Ross Goroshin, Yutian Chen, Simon Osindero, Joao Carreira, Razvan Pascanu
- Accepted by TMLR
- Readers: Everyone
Bridging Graph Neural Networks and Large Language Models: A Survey and Unified Perspective
Alvaro Arroyo, Federico Barbero, Hugh Blayney, Michael M. Bronstein, Xiaowen Dong, Pietro Lio, Razvan Pascanu, Pierre Vandergheynst
- OpenReview Preprint
- Readers: Everyone
Fine-Tuned In-Context Learners
Jörg Bornschein, Amal Rannen-Triki, Razvan Pascanu, Yazhe Li, Clare Lyle, Xu Owen He
- Submitted to ICLR 2026
- Readers: Everyone
LLMs are Greedy Agents: Effects of RL Fine-tuning on Decision-Making Abilities
Thomas Schmied, Jörg Bornschein, Jordi Grau-Moya, Markus Wulfmeier, Razvan Pascanu
- ICLR 2026 Poster
- Readers: Everyone
MesaNet: Sequence Modeling by Locally Optimal Test-Time Training
Johannes von Oswald, Nino Scherrer, Seijin Kobayashi, Luca Versari, Songlin Yang, Maximilian Schlegel, Kaitlin Maile, Yanick Schimpf, Oliver Sieberling, Alexander Meulemans, Guillaume Lajoie, Rif A. Saurous, Charlotte Frenkel, Razvan Pascanu, Blaise Aguera y Arcas, Joao Sacramento
- ICLR 2026 Poster
- Readers: Everyone
From Markov to Laplace: How Mamba In-Context Learns Markov Chains
Marco Bondaschi, Nived Rajaraman, Xiuying Wei, Razvan Pascanu, Caglar Gulcehre, Michael Gastpar, Ashok Vardhan Makkuva
- ICLR 2026 Oral
- Readers: Everyone
Lattice: Learning to Efficiently Compress the Memory
Mahdi Karami, Razvan Pascanu, Vahab Mirrokni
- Submitted to ICLR 2026
- Readers: Everyone

View all 113 publications

Co-Authors

View all 321 co-authors