ResearchBenchmarksAverage Reward on AIM (N=200, T=10, K=30)Average Reward on AIM (N=200, T=10, K=30)Follow137.28Average RewardLMTA128.284130.6195132.955135.2905May 16, 2026Evaluation resultsMethodMethodLinksAverage RewardLMTA2026.05137.28WS-Option2026.05128.63