With the change `env = create_env(env_name, env_configs[env_name])` in https://github.com/NVIDIA-NeMo/RL/pull/2056, eval could support for different environments. But currently MathMultiRewardEnvironment is not supported for eval. https://github.com/NVIDIA-NeMo/RL/pull/2069#discussion_r2903293464
With the change
env = create_env(env_name, env_configs[env_name])in #2056, eval could support for different environments.But currently MathMultiRewardEnvironment is not supported for eval. #2069 (comment)