diff --git a/test.py b/test.py index 70f53d6..12e3936 100755 --- a/test.py +++ b/test.py @@ -27,9 +27,10 @@ def main(env_name='ColumbusCandyland_Aux10-v0', timesteps=1_000_000, showRes=Tru ppo = PPO( MlpPolicyPPO, env, - projection=BaseProjectionLayer(), # KLProjectionLayer(trust_region_coeff=0.01), - policy_kwargs={'dist_kwargs': {'neural_strength': Strength.NONE, 'cov_strength': Strength.FULL, 'parameterization_type': - ParametrizationType.CHOL, 'enforce_positive_type': EnforcePositiveType.ABS, 'prob_squashing_type': ProbSquashingType.NONE}}, + # KLProjectionLayer(trust_region_coeff=0.01), + projection=KLProjectionLayer(trust_region_coeff=0.01), + policy_kwargs={'dist_kwargs': {'neural_strength': Strength.NONE, 'cov_strength': Strength.DIAG, 'parameterization_type': + ParametrizationType.NONE, 'enforce_positive_type': EnforcePositiveType.ABS, 'prob_squashing_type': ProbSquashingType.NONE}}, verbose=0, tensorboard_log=root_path+"/logs_tb/" + env_name+"/ppo"+(['', '_sde'][use_sde])+"/",