ResearchBenchmarksG2 balanced on Hopper Medium-Replay v2Follow781.3Mean Raw MuJoCo ReturnCQL11.908211.654411.4611.146Apr 23, 2026Evaluation ResultsMethodMethodLinksMean Raw MuJoCo ReturnCQL2026.04781.3PoEalpha=0.9alpha=0.92026.04460.6Frozen2026.04456.1IQL2026.04456PoEalpha=0.5alpha=0.52026.04436.4KL-Regalpha=0.5alpha=0.52026.04436.4Additivealpha=0.5alpha=0.52026.04176AWR2026.04157.9Prior2026.0441.5