scalable training
The ability to efficiently train AI models on increasingly large datasets without a corresponding exponential increase in computational resources or time.
- Audio-Sync Video Generation with Multi-Stream Temporal Control
- Consistently Simulating Human Personas with Multi-Turn Reinforcement Learning
- Multi-Agent Imitation by Learning and Sampling from Factorized Soft Q-Function
- Non-equilibrium Annealed Adjoint Sampler
- PLANA3R: Zero-shot Metric Planar 3D Reconstruction via Feed-forward Planar Splatting
- SCAN: Self-Denoising Monte Carlo Annotation for Robust Process Reward Learning
- Through the River: Understanding the Benefit of Schedule-Free Methods for Language Model Training
- ToolRL: Reward is All Tool Learning Needs