Scholar
Luke Marris
Google Scholar ID: dvTeSX4AAAAJ
Research Engineer at DeepMind, PhD from University College London
Citations & Impact
Citations
3,199
H-index
13
i10-index
15
Publications
20
Co-authors
24
list available
Contact
Publications
Resume (English only)
Academic Achievements
- 1. Gemini 2.5: Pushing the Frontier with Advanced Reasoning, Multimodality, Long Context, and Next Generation Agentic Capabilities, 2025, arXiv 2025.
- 2. Re-evaluating Open-ended Evaluation of Large Language Models, 2025, arXiv 2025.
- 3. Deviation Ratings: A General, Clone-Invariant Rating Method, 2025, arXiv 2025, GAIW 2025.
Research Experience
- Staff Research Engineer, Google DeepMind, London.
Education
- PhD, University College London, Thesis: Multiagent Training in N-Player General-Sum Games; Information Engineering, Bachelors and Masters, First Class, University of Cambridge.
Background
- An artificial intelligence engineer and researcher. Expertise in machine learning, optimization, deep learning, reinforcement learning, game theory, and multiagent systems. Particularly interested in training deep reinforcement agents at scale in many-player mixed-motive games, with a focus on building principled learning algorithms that provably select and compute equilibria.
Miscellany
- Developed the twoxtwogame LaTeX package for the visualization of 2x2 normal form games. The package is based on PGF/TikZ and produces beautiful vector graphics intended for use in scientific publications.
