AI Paper Atlas

Research map / Reinforcement learning / Visual and representation RL

Scaling Agent Learning via Experience Synthesis

Zhaorun Chen, Zhuokai Zhao, Kai Zhang, Bo Liu, Qi Qi, Yifan Wu, Tarun Kalluri, Xuefei Cao, Yuanhao Xiong, Haibo Tong, Huaxiu Yao, Hengduo Li and 6 more

ICLR 2026 · Reinforcement learning › Visual and representation RL · approach: Scalable RL systems and environments · 47 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 experience replay (ablation)DreamGym · WebShopsuccess rate (%): 59.2 -> 63.9Improved
add experience replay (ablation)DreamGym · WebArenasuccess rate (%): 9.7 -> 13.3Improved
add experience reasoning (ablation)DreamGym · WebShopsuccess rate (%): 55.8 -> 63.9Improved
add experience reasoning (ablation)DreamGym · WebArenasuccess rate (%): 7.3 -> 13.3Improved
add curriculum task generation (ablation)DreamGym · WebShopsuccess rate (%): 57.3 -> 63.9Improved
add curriculum task generation (ablation)DreamGym · WebArenasuccess rate (%): 7.3 -> 13.3Improved
train on synthetic experience only instead of 80K real transitions (GRPO)Qwen2.5-7B GRPO · WebArenasuccess rate (%): 6.1 -> 12.7Improved
train on synthetic experience only instead of 80K real transitions (GRPO)Qwen2.5-7B GRPO · ALFWorldsuccess rate (%): 79.8 -> 71.0Hurt
add sim-to-real RL with 5K real transitions after synthetic training (GRPO)Llama-3.2-3B GRPO · WebShopsuccess rate (%): 59.3 -> 70.5Improved
add sim-to-real RL with 5K real transitions after synthetic training (PPO)Llama-3.1-8B PPO · WebShopsuccess rate (%): 58.1 -> 63.9Improved
transfer DreamGym policy from WebShop/WebArena to ALFWorldDreamGym policy · ALFWorldsuccess rateHurt

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