Reward Functions¶
Similar structures are provided with the action and state space. Meta-World provides well-shaped reward functions for the individual tasks that are solvable by current single-task reinforcement learning approaches. To assure equivalent learning in the settings with multiple tasks, all task rewards have the same magnitude.
Options¶
Meta-World currently implements two types of reward functions that can be selected
by passing the reward_func_version
keyword argument to gym.make(...)
.
Version 1¶
Passing reward_func_version=v1
configures the benchmark with the original
reward function of Meta-World, which is actually a version of the
pick-place-wall
task that is modified to also work for the other tasks. Any paper that reports results on the ‘env-name-v1’ environments, uses this reward function.
Version 2¶
Passing reward_func_version=v2
configures the benchmark with the updated
reward functions of Meta-World. Any paper that reports results on the ‘env-name-v2’ environments, uses this reward function.