AI Paper Atlas

Research map / Reinforcement learning / RL from feedback

Gradient-Aware Scheduling: Coupling Curriculum and Staleness for Async Reinforcement Learning

Xinyu Zhang

ICML 2026 · Reinforcement learning › RL from feedback · approach: Scaling RL with verifiable rewards

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 curriculum-staleness coupling (CSC) (ablation)GAS · Code (HumanEval + synthetic)Pass@1 (%): 42.1 -> 60.1 (-18.0 without CSC)Improved
add execution-aware staleness budgets (EAAS) (ablation)GAS · Code (HumanEval + synthetic)Pass@1 (%): 55.3 -> 60.1 (-4.8 without EAAS)Improved
add execution-aware staleness budgets (EAAS) (ablation)GAS · Code (HumanEval + synthetic)throughput (samples/s): 16.8 -> 22.4Improved
add bandit curriculum (ACB) (ablation)GAS · Code (HumanEval + synthetic)Pass@1 (%): 46.9 -> 60.1 (-13.2 without ACB)Improved
add CSC (discarding stale experiences) (ablation)GAS · Code (HumanEval + synthetic)throughput (samples/s): 23.5 -> 22.4Hurt
use asynchronous GRPO instead of synchronousGRPO · Code (HumanEval + synthetic)Pass@1 (%): 39.7 -> 31.8 Pass@1; 9.7 -> 24.3 throughputMixed
add uniform staleness control to async GRPOAsync-GRPO · Code (HumanEval + synthetic)Pass@1 (%): 31.8 -> 40.3Improved
add difficulty-aware coupling instead of uniform staleness controlAsync-GRPO · Code (HumanEval + synthetic)Pass@1 (%): 40.3 -> 60.1Improved
add curriculum (CCCS) to synchronous GRPOSync-GRPO · Code (HumanEval + synthetic)Pass@1 (%): 39.7 -> 51.5Improved
scale GAS to Qwen3.5-9B vs sync baselineGAS · HumanEval (Qwen3.5-9B)training reward: 0.279 -> 0.430Improved
GAS vs plain async GRPO at 9BGAS · HumanEval (Qwen3.5-9B)training reward: 0.196 -> 0.430Improved
transfer GAS to GSM8K with same hyperparametersGAS · GSM8K (Qwen3.5-9B)training reward: 0.312 -> 0.572Improved

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