0ce29ef16326b0388142b8e51fe659abd6388311,examples/her/her_sac_gym_fetch_reach.py,,experiment,#,14

Before Change

def experiment(variant):
    eval_env = gym.make("FetchReach-v1")
    expl_env = gym.make("FetchReach-v1")

    observation_key = "observation"
    desired_goal_key = "desired_goal"

    achieved_goal_key = desired_goal_key.replace("desired", "achieved")
    replay_buffer = ObsDictRelabelingBuffer(
        env=eval_env,
        observation_key=observation_key,
        desired_goal_key=desired_goal_key,

After Change

def experiment(variant):
    // unwrap the TimeLimitEnv wrapper since we manually termiante after 50 steps
    eval_env = gym.make("FetchReach-v1").env
    expl_env = gym.make("FetchReach-v1").env

    observation_key = "observation"
    desired_goal_key = "desired_goal"

    achieved_goal_key = desired_goal_key.replace("desired", "achieved")
    replay_buffer = ObsDictRelabelingBuffer(
        env=eval_env,
        observation_key=observation_key,
        desired_goal_key=desired_goal_key,

Instances