More regularization

Author: RunnersNum40

experiments/g1/main.py

@@ -36,12 +36,12 @@ class ExperimentConfig:
     torso_offset_range: tuple[float, float] = (-2.0, 5.0)
 
     # L2C2
-    l2c2_lambda_pi: float = 1.0
-    l2c2_lambda_v: float = 0.1
+    l2c2_lambda_pi: float = 5.0
+    l2c2_lambda_v: float = 0.5
 
     # Reward weights
     task_weight: float = 2.5
-    regularization_weight: float = 0.1
+    regularization_weight: float = 1.0
     style_weight: float = 1.0
     target_weight: float = 1.0