Steven Kapturowski
Papers
1
Total Citations
2
H-Index
1
About
Steven Kapturowski is a leading researcher at the intersection of deep reinforcement learning (RL) and large-scale model training. His primary contributions lie in advancing offline RL algorithms, particularly demonstrating that actor-critic methods can effectively scale to massive architectures like transformers. His most cited work, "Offline Actor-Critic Reinforcement Learning Scales to Large Models" (2024), provides a pivotal proof-of-concept that offline RL follows similar scaling laws as supervised learning, outperforming strong behavioral cloning baselines in multi-task settings. This finding challenges conventional wisdom and opens new pathways for training large models without online interaction. While his citation count is still growing, the conceptual impact of this work is significant, positioning Kapturowski as a key voice in the future of scalable, data-driven decision-making. His research bridges the gap between traditional RL and modern large-scale machine learning, offering practical insights for researchers aiming to deploy RL in real-world, offline scenarios.
Research Focus
Key Achievements
Top Papers
- 1Offline Actor-Critic Reinforcement Learning Scales to Large Models2 citations · 2024