AI Paper Atlas

Research map / Reinforcement learning / Visual and representation RL

RAD-2: Scaling Reinforcement Learning in a Generator-Discriminator Framework

Hao Gao, Shaoyu Chen, Yifan Zhu, Yuehao Song, Wenyu Liu, Qian W. L. Zhang, Xinggang Wang

TPAMI 2026 · Reinforcement learning › Visual and representation RL · approach: RL for combinatorial routing · 6 citations (OpenAlex)

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
add on-policy generator RL optimization (ablation)IL-pretrained diffusion planner · BEV-Warp safety-oriented (512 scenarios)CR: 0.533 -> 0.287Improved
add on-policy generator RL optimization (ablation)IL-pretrained diffusion planner · BEV-Warp efficiency-oriented (512 scenarios)EP@1.0: 0.516 -> 0.391Hurt
add IL fine-tuning stage during generator optimization (ablation)generator RL · BEV-Warp efficiency-oriented (512 scenarios)EP@1.0: 0.391 -> 0.527Improved
add RL discriminator training alone (ablation)IL-pretrained planner · BEV-Warp safety-oriented (512 scenarios)CR: 0.533 -> 0.337Improved
add discriminator RL to generator RL + IL fine-tuning (full RAD-2) (ablation)RAD-2 · BEV-Warp safety-oriented (512 scenarios)CR: 0.403 -> 0.234Improved
initialize discriminator from planning head instead of random (ablation)RAD-2 discriminator · BEV-Warp safety-oriented (512 scenarios)CR: 0.426 -> 0.337Improved
add reward-variance clip filtering (ablation)RAD-2 · BEV-Warp efficiency-oriented (512 scenarios)EP@1.0: 0.662 -> 0.728Improved
add reward-variance clip filtering (ablation)RAD-2 · BEV-Warp safety-oriented (512 scenarios)CR: 0.340 -> 0.337No clear effect
add entropy regularization to RL objective (ablation)RAD-2 · BEV-Warp safety-oriented (512 scenarios)CR: 0.254 -> 0.234Improved
execution horizon H_reuse 8 vs 2 (ablation)RAD-2 · BEV-Warp safety-oriented (512 scenarios)CR: 0.355 -> 0.337Mixed
TC-GRPO group size 4 vs 2 and 8 (ablation)RAD-2 · BEV-Warp safety-oriented (512 scenarios)CR: 0.313 (2), 0.234 (4), 0.291 (8)Improved
increase inference candidates M from 8 to 128 (ablation)RAD-2 · BEV-Warp efficiency-oriented (512 scenarios)EP@1.0: 0.667 -> 0.814Improved

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