AI Paper Atlas

Research map / Reinforcement learning / Visual and representation RL

CompleteP for RL: Maintaining Feature Learning When Scaling Deep Reinforcement Learning

Adam Lee, M Ganesh Kumar, Blake Bordelon, Cengiz Pehlevan

ICML 2026 · Reinforcement learning › Visual and representation RL · approach: Scalable RL systems and environments

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use CompleteP instead of NTK parameterizationCompleteP · G1JoystickRoughTerrain (L=2)compute to reach same reward: -31%Improved
use CompleteP instead of NTK parameterizationCompleteP · G1JoystickRoughTerrain (L=2)reward at equal compute: +33%Improved
use CompleteP instead of NTK parameterizationCompleteP · AcrobotSwingupSparse (L=1)reward at equal compute: +701%Improved
use CompleteP instead of NTK parameterizationCompleteP · CartpoleSwingup (L=2)reward at equal compute: +1%No clear effect
use CompleteP instead of NTK parameterizationCompleteP · humanoid_u_maze (L=4)reward at equal compute: +27%Improved
scale width and depth with CompleteP (constant learning rate)CompleteP · HalfCheetah, HumanoidMazefinal reward vs learning rateImproved
scale depth with NTK without lowering learning rateNTK · HalfCheetah, HumanoidMazefinal rewardHurt
increase width from N=64 to N=2048 under NTKNTK · reward vs env stepslearning speedHurt
use CompleteP instead of NTK for width scalingCompleteP · feature kernel across seeds (CheetahRun)CKA across seedsImproved
use CompleteP instead of NTKCompleteP · logit evolution rate vs widthMSE of logitsNo clear effect
add orthogonal initialization or layer normalization to NTKNTK · reward at lower computemaximal rewardMixed
use CompleteP instead of NTK with sparse rewardsCompleteP · sparse-reward tasksefficiencyImproved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy