Upload New File

critic_gradients.m

+function gradients = critic_gradients(network, input_data1, input_data2,target_Q_values) 
+    lambda = 0.99;
+    predictions = predict(network, input_data1, input_data2);
+    target_Q_values = reshape(target_Q_values,length(predictions),[]);
+    loss = lambda*crossentropy(dlarray((mean(target_Q_values,2)),'BC'),predictions);
+    loss = real(sum(loss));
+    gradients = dlgradient(-loss, network.Learnables);
+end
\ No newline at end of file