update Walk.py and update model of Walk version 0.2.0

scripts/gyms/Walk.py

@@ -247,7 +247,7 @@ class WalkEnv(gym.Env):
         self.reward_forward_lean_penalty_cap = 0.7
         self.reward_knee_straight_threshold = 0.18
         self.reward_knee_straight_penalty_scale = 0.70
-        self.reward_hip_overextend_threshold = 0.95
+        self.reward_hip_overextend_threshold = 1.1
         self.reward_hip_overextend_penalty_scale = 1.30
         self.reward_leg_stretch_penalty_scale = 1.20
         self.reward_stretch_lean_combo_scale = 1.40
@@ -886,7 +886,7 @@ class Train(Train_Base):
         server_warmup_sec = 3.0
         n_steps_per_env = 1024  # RolloutBuffer is of size (n_steps_per_env * n_envs)
         minibatch_size = 1024  # should be a factor of (n_steps_per_env * n_envs)
-        total_steps = 18000000
+        total_steps = 180000000
         learning_rate = 3e-4
         ent_coef = 0.05
         clip_range = 0.2