Search this site
Embedded Files
Skip to main content
Skip to navigation
ICML 2025 supplement
Value-Based RL Scales Predictably
Statistical analysis of budget plots.Â
Extrapolation results for the Pareto frontier on OpenAI gym
Consistent budget extrapolation fits
Pareto frontier extrapolation to UTD 64
Data efficiency fits (Eq 4.1) on OpenAI gym - Multiple values for J
Hyperparameter selection - quantitative experiments
Figure 1 (left) - ticks set to only be power of 2
Figure 1 (left) with value of J explicitly stated
Isotonic regression example: using gaussian smoothing with sigma=3 leads to both oversmoothing (right) and undersmoothing (left)
Detailed hyperaparameters
Google Sites
Report abuse
Page details
Page updated
Google Sites
Report abuse