Aviral Kumar
Source-listed: Assistant Professor · Carnegie Mellon University, School of Computer Science
Record snapshot: Sep 3, 2026. Coverage and affiliations may be incomplete or historical.
Selected work
5- Training Language Models to Self-Correct via Reinforcement LearningSaved credit: Contributor · 2024Explore people connected to this work →
- Scaling LLM Test-Time Compute Optimally can be More Effective than Scaling Model ParametersSaved credit: Contributor · 2024Explore people connected to this work →
- DigiRL: Training In-The-Wild Device-Control Agents with Autonomous Reinforcement LearningSaved credit: Contributor · 2024Explore people connected to this work →
- Conservative Q-Learning for Offline Reinforcement LearningSaved credit: Contributor · 2020Explore people connected to this work →
- Gemini: A Family of Highly Capable Multimodal ModelsSaved credit: Contributor · 2023Explore people connected to this work →