ResearchBenchmarksLocomotion on Antmaze m-play v2Follow89.5D4RL Score (%)ReBRAC87.73288.19188.6589.109May 22, 2026Evaluation ResultsMethodMethodLinksD4RL Score (%)Time per Environment Step (ms)ReBRACEpisodes=100, Seeds=4Episodes=100, Seeds=42026.0589.5—InverterEpisodes=100, Seeds=4,...Episodes=100, Seeds=4, Hardware=single A40 GPU, Framework=PyTorch, Batch size=1, Cuda-synced=true2026.0587.80.13BC-10%Hardware=single A40 GP...Hardware=single A40 GPU, Framework=PyTorch, Batch size=1, Cuda-synced=true2026.05—1.25