diff --git a/examples/VisualAutoencoder/uvisualautoencodertinyimagenet.pas b/examples/VisualAutoencoder/uvisualautoencodertinyimagenet.pas index 92175aaa..0d0b5d0e 100644 --- a/examples/VisualAutoencoder/uvisualautoencodertinyimagenet.pas +++ b/examples/VisualAutoencoder/uvisualautoencodertinyimagenet.pas @@ -249,8 +249,8 @@ procedure TFormVisualLearning.Learn( Sender: TObject); TNNetConvolution.Create({Features=}32 * NeuronMultiplier,{FeatureSize=}3,{Padding=}1,{Stride=}1,{SuppressBias=}1), TNNetConvolution.Create({Features=}32 * NeuronMultiplier,{FeatureSize=}3,{Padding=}1,{Stride=}2,{SuppressBias=}1), //16x16 TNNetConvolution.Create({Features=}32 * NeuronMultiplier,{FeatureSize=}3,{Padding=}1,{Stride=}1,{SuppressBias=}1), - TNNetConvolution.Create({Features=}64 * NeuronMultiplier,{FeatureSize=}3,{Padding=}1,{Stride=}2,{SuppressBias=}0), //8x8 - TNNetConvolution.Create({Features=}64 * NeuronMultiplier,{FeatureSize=}3,{Padding=}1,{Stride=}1,{SuppressBias=}0), + TNNetConvolution.Create({Features=}64 * NeuronMultiplier,{FeatureSize=}3,{Padding=}1,{Stride=}2,{SuppressBias=}1), //8x8 + TNNetConvolution.Create({Features=}64 * NeuronMultiplier,{FeatureSize=}3,{Padding=}1,{Stride=}1,{SuppressBias=}1), TNNetConvolution.Create({Features=}128 * NeuronMultiplier,{FeatureSize=}3,{Padding=}1,{Stride=}2,{SuppressBias=}1), //4x4 TNNetConvolution.Create({Features=}128 * NeuronMultiplier,{FeatureSize=}3,{Padding=}1,{Stride=}1,{SuppressBias=}1), @@ -276,8 +276,6 @@ procedure TFormVisualLearning.Learn( Sender: TObject); FAutoencoder.LoadFromFile(FBaseName+'autoencoder.nn'); end; FAutoencoder.DebugStructure(); - FAutoencoder.SetLearningRate(0.001,0.9); - FAutoencoder.SetL2Decay(0.0); FFit.OnAfterEpoch := @Self.AutoencoderOnAfterEpoch; FFit.OnAfterStep := @Self.AutoencoderOnAfterStep; @@ -296,7 +294,7 @@ procedure TFormVisualLearning.Learn( Sender: TObject); FAutoencoder.EnableOpenCL(FEasyOpenCL.PlatformIds[0], FEasyOpenCL.Devices[0]); end; {$endif} - //Debug only: FFit.MaxThreadNum := 2; + //Debug only: FFit.MaxThreadNum := 1; FFit.FitLoading(FAutoencoder, {EpochSize=}FTrainImages.CountElements(), 0, 0, {Batch=}64, {Epochs=}35000, @GetTrainingData, nil, nil); // This line does the same as above FAutoencoder.Free;