Scholar
Nathan Grinsztajn
Google Scholar ID: yVHIYEYAAAAJ
Cohere
Citations & Impact
Citations
379
H-index
10
i10-index
10
Publications
20
Co-authors
0
Publications
Cited
4
Resume (English only)
Academic Achievements
- - Paper: Combinatorial Optimization with Policy Adaptation using Latent Space Search
- - Paper: Winner Takes It All: Training Performant RL Populations for Combinatorial Optimization
- - Paper: More Efficient Exploration with Symbolic Priors on Action Sequence Equivalences
- - Paper: MetaREVEAL: RL-based Meta-learning from Learning Curves
- - Paper: There Is No Turning Back: A Self-Supervised Approach for Reversibility-Aware Reinforcement Learning
Research Experience
- - Technical Staff, 2024-Present, Cohere
- - Research Scientist, 2023-2024, InstaDeep
- - Research Intern, April 2022 - October 2022, InstaDeep, supervised by Thomas D. Barrett
- - PhD Student, October 2019 - April 2023, Inria, supervised by P. Preux
- - Graduate Research Intern, April 2018 - August 2018, UC Berkeley, supervised by S. Dudoit
- - Blockchain Developer (intern), June 2017 - November 2017, BitSpread Ltd
Education
- Ph.D. in reinforcement learning for combinatorial optimization from Inria/CNRS, 2019-2023, under the supervision of P. Preux.
Background
- Interests: Reinforcement Learning, Large Language Models, Combinatorial Optimization. Biography: Works on RL, LLMs, and their interactions at Cohere. Previously, a research scientist at InstaDeep, focusing on using transformers for combinatorial optimization and discrete problems.
Co-authors