NeurIPS 2025 Explorer
Concepts
Authors
Glossary
johnsanterre.github.io
update-to-data ratio
A measure of the frequency and extent of model updates in relation to incoming data. Balancing the update-to-data ratio is crucial for optimizing model performance and stability over time.
3 papers
Compute-Optimal Scaling for Value-Based Deep RL
Scaling Off-Policy Reinforcement Learning with Batch and Weight Normalization
Staggered Environment Resets Improve Massively Parallel On-Policy Reinforcement Learning