save last training hyper param.

MADEAPPS · Sep 20, 2024 · b097892 · b097892
1 parent dcfe189
commit b097892
Show file tree

Hide file tree

Showing 2 changed files with 33 additions and 21 deletions.
diff --git a/newton-4.00/applications/ndSandbox/demos/ndAdvancedIndustrialRobot.cpp b/newton-4.00/applications/ndSandbox/demos/ndAdvancedIndustrialRobot.cpp
@@ -560,23 +560,35 @@ namespace ndAdvancedRobot
 			ndFloat32 rewardWeigh = 1.0f / 6.0f;
 			ndFloat32 azimuthReward = ScalarReward(positError2.m_z);
 
-			ndFloat32 reward = rewardWeigh * azimuthReward;
-			if (azimuthReward > 0.5f)
-			{
-				const ndVector rotationError(CalculateDeltaTargetRotation(currentEffectorMatrix));
-				const ndVector rotationError2 = rotationError * rotationError;
-
-				ndFloat32 omega_xReward = rewardWeigh * GaussianReward(rotationError2.m_x);
-				ndFloat32 omega_yReward = rewardWeigh * GaussianReward(rotationError2.m_y);
-				ndFloat32 omega_zReward = rewardWeigh * GaussianReward(rotationError2.m_z);
-				reward += (omega_xReward + omega_yReward + omega_zReward);
-				if ((omega_xReward > 1.0e-3f) || (omega_yReward > 1.0e-3f) || (omega_zReward > 1.0e-3f))
-				{
-					ndFloat32 posit_xReward = rewardWeigh * GaussianReward(positError2.m_x);
-					ndFloat32 posit_yReward = rewardWeigh * GaussianReward(positError2.m_y);
-					reward += (posit_xReward + posit_yReward);
-				}
-			}
+			//ndFloat32 reward = rewardWeigh * azimuthReward;
+			//f (azimuthReward > 0.5f)
+			//
+			//	const ndVector rotationError(CalculateDeltaTargetRotation(currentEffectorMatrix));
+			//	const ndVector rotationError2 = rotationError * rotationError;
+			//
+			//	ndFloat32 omega_xReward = rewardWeigh * GaussianReward(rotationError2.m_x);
+			//	ndFloat32 omega_yReward = rewardWeigh * GaussianReward(rotationError2.m_y);
+			//	ndFloat32 omega_zReward = rewardWeigh * GaussianReward(rotationError2.m_z);
+			//	reward += (omega_xReward + omega_yReward + omega_zReward);
+			//	if ((omega_xReward > 1.0e-3f) || (omega_yReward > 1.0e-3f) || (omega_zReward > 1.0e-3f))
+			//	{
+			//		ndFloat32 posit_xReward = rewardWeigh * GaussianReward(positError2.m_x);
+			//		ndFloat32 posit_yReward = rewardWeigh * GaussianReward(positError2.m_y);
+			//		reward += (posit_xReward + posit_yReward);
+			//	}
+
+			ndFloat32 posit_xReward = GaussianReward(positError2.m_x);
+			ndFloat32 posit_yReward = GaussianReward(positError2.m_y);
+
+			const ndVector rotationError(CalculateDeltaTargetRotation(currentEffectorMatrix));
+			const ndVector rotationError2 = rotationError * rotationError;
+
+			ndFloat32 omega_xReward = GaussianReward(rotationError2.m_x);
+			ndFloat32 omega_yReward = GaussianReward(rotationError2.m_y);
+			ndFloat32 omega_zReward = GaussianReward(rotationError2.m_z);
+
+			ndFloat32 reward = (azimuthReward + posit_xReward + posit_yReward + 
+							    omega_xReward + omega_yReward + omega_zReward) * rewardWeigh;
 			return reward;
 		}
 

diff --git a/newton-4.00/sdk/dBrain/ndBrainAgentContinuePolicyGradient_Trainer.cpp b/newton-4.00/sdk/dBrain/ndBrainAgentContinuePolicyGradient_Trainer.cpp
@@ -50,10 +50,10 @@ ndBrainAgentContinuePolicyGradient_TrainerMaster::HyperParameters::HyperParamete
 
 	//m_criticLearnRate = ndBrainFloat(0.0004f);
 	//m_policyLearnRate = ndBrainFloat(0.0002f);
-	//m_criticLearnRate = ndBrainFloat(0.0002f);
-	//m_policyLearnRate = ndBrainFloat(0.0001f);
-	m_criticLearnRate = ndBrainFloat(0.0005f);
-	m_policyLearnRate = ndBrainFloat(0.001f);
+	m_criticLearnRate = ndBrainFloat(0.0001f);
+	m_policyLearnRate = ndBrainFloat(0.0002f);
+	//m_criticLearnRate = ndBrainFloat(0.0005f);
+	//m_policyLearnRate = ndBrainFloat(0.001f);
 
 	m_regularizer = ndBrainFloat(1.0e-6f);
 	m_discountFactor = ndBrainFloat(0.99f);