---------------------------------------------------------------------- [Sequential] -------------------------------------------------------------------- [LoweringConvolution] filter size : (3, 3) input shape : {32, 32, 3} output shape : {30, 30, 64} ------------------------------------------------------------------ [ConvolutionIm2Col] input shape : {32, 32, 3} output shape : {3, 3, 3} ------------------------------------------------------------------ [Sequential] ---------------------------------------------------------------- [StochasticLutBn] -------------------------------------------------------------- [StochasticBatchNormalization] input shape : {3, 3, 3} output shape : {3, 3, 3} -------------------------------------------------------------- [StochasticLut6] input shape : {3, 3, 3} output shape : {512} ---------------------------------------------------------------- [StochasticLutBn] -------------------------------------------------------------- [StochasticBatchNormalization] input shape : {512} output shape : {512} -------------------------------------------------------------- [StochasticLut6] input shape : {512} output shape : {384} ---------------------------------------------------------------- [StochasticLutBn] -------------------------------------------------------------- [StochasticBatchNormalization] input shape : {384} output shape : {384} -------------------------------------------------------------- [StochasticLut6] input shape : {384} output shape : {64} ------------------------------------------------------------------ [ConvolutionCol2Im] input shape : {64} output shape : {30, 30, 64} -------------------------------------------------------------------- [LoweringConvolution] filter size : (3, 3) input shape : {30, 30, 64} output shape : {28, 28, 64} ------------------------------------------------------------------ [ConvolutionIm2Col] input shape : {30, 30, 64} output shape : {3, 3, 64} ------------------------------------------------------------------ [Sequential] ---------------------------------------------------------------- [StochasticLutBn] -------------------------------------------------------------- [StochasticBatchNormalization] input shape : {3, 3, 64} output shape : {3, 3, 64} -------------------------------------------------------------- [StochasticLut6] input shape : {3, 3, 64} output shape : {512} ---------------------------------------------------------------- [StochasticLutBn] -------------------------------------------------------------- [StochasticBatchNormalization] input shape : {512} output shape : {512} -------------------------------------------------------------- [StochasticLut6] input shape : {512} output shape : {384} ---------------------------------------------------------------- [StochasticLutBn] -------------------------------------------------------------- [StochasticBatchNormalization] input shape : {384} output shape : {384} -------------------------------------------------------------- [StochasticLut6] input shape : {384} output shape : {64} ------------------------------------------------------------------ [ConvolutionCol2Im] input shape : {64} output shape : {28, 28, 64} -------------------------------------------------------------------- [MaxPooling] filter size : (2, 2) input shape : {28, 28, 64} output shape : {14, 14, 64} -------------------------------------------------------------------- [LoweringConvolution] filter size : (3, 3) input shape : {14, 14, 64} output shape : {12, 12, 128} ------------------------------------------------------------------ [ConvolutionIm2Col] input shape : {14, 14, 64} output shape : {3, 3, 64} ------------------------------------------------------------------ [Sequential] ---------------------------------------------------------------- [StochasticLutBn] -------------------------------------------------------------- [StochasticBatchNormalization] input shape : {3, 3, 64} output shape : {3, 3, 64} -------------------------------------------------------------- [StochasticLut6] input shape : {3, 3, 64} output shape : {1024} ---------------------------------------------------------------- [StochasticLutBn] -------------------------------------------------------------- [StochasticBatchNormalization] input shape : {1024} output shape : {1024} -------------------------------------------------------------- [StochasticLut6] input shape : {1024} output shape : {768} ---------------------------------------------------------------- [StochasticLutBn] -------------------------------------------------------------- [StochasticBatchNormalization] input shape : {768} output shape : {768} -------------------------------------------------------------- [StochasticLut6] input shape : {768} output shape : {128} ------------------------------------------------------------------ [ConvolutionCol2Im] input shape : {128} output shape : {12, 12, 128} -------------------------------------------------------------------- [LoweringConvolution] filter size : (3, 3) input shape : {12, 12, 128} output shape : {10, 10, 64} ------------------------------------------------------------------ [ConvolutionIm2Col] input shape : {12, 12, 128} output shape : {3, 3, 128} ------------------------------------------------------------------ [Sequential] ---------------------------------------------------------------- [StochasticLutBn] -------------------------------------------------------------- [StochasticBatchNormalization] input shape : {3, 3, 128} output shape : {3, 3, 128} -------------------------------------------------------------- [StochasticLut6] input shape : {3, 3, 128} output shape : {1024} ---------------------------------------------------------------- [StochasticLutBn] -------------------------------------------------------------- [StochasticBatchNormalization] input shape : {1024} output shape : {1024} -------------------------------------------------------------- [StochasticLut6] input shape : {1024} output shape : {768} ---------------------------------------------------------------- [StochasticLutBn] -------------------------------------------------------------- [StochasticBatchNormalization] input shape : {768} output shape : {768} -------------------------------------------------------------- [StochasticLut6] input shape : {768} output shape : {64} ------------------------------------------------------------------ [ConvolutionCol2Im] input shape : {64} output shape : {10, 10, 64} -------------------------------------------------------------------- [MaxPooling] filter size : (2, 2) input shape : {10, 10, 64} output shape : {5, 5, 64} -------------------------------------------------------------------- [StochasticLutBn] ------------------------------------------------------------------ [StochasticBatchNormalization] input shape : {5, 5, 64} output shape : {5, 5, 64} ------------------------------------------------------------------ [StochasticLut6] input shape : {5, 5, 64} output shape : {2048} -------------------------------------------------------------------- [StochasticLutBn] ------------------------------------------------------------------ [StochasticBatchNormalization] input shape : {2048} output shape : {2048} ------------------------------------------------------------------ [StochasticLut6] input shape : {2048} output shape : {1024} -------------------------------------------------------------------- [StochasticLutBn] ------------------------------------------------------------------ [StochasticBatchNormalization] input shape : {1024} output shape : {1024} ------------------------------------------------------------------ [StochasticLut6] input shape : {1024} output shape : {360} -------------------------------------------------------------------- [StochasticLutBn] ------------------------------------------------------------------ [StochasticBatchNormalization] input shape : {360} output shape : {360} ------------------------------------------------------------------ [StochasticLut6] input shape : {360} output shape : {60} -------------------------------------------------------------------- [StochasticLutBn] ------------------------------------------------------------------ [StochasticBatchNormalization] input shape : {60} output shape : {60} ------------------------------------------------------------------ [StochasticLut6] input shape : {60} output shape : {10} ---------------------------------------------------------------------- fitting start : Cifar10Sparse6Cnn 179.49s epoch[ 1] test accuracy : 0.5921 train accuracy : 0.6058 421.43s epoch[ 2] test accuracy : 0.6853 train accuracy : 0.7112 664.79s epoch[ 3] test accuracy : 0.7130 train accuracy : 0.7519 908.20s epoch[ 4] test accuracy : 0.7247 train accuracy : 0.7650 1151.73s epoch[ 5] test accuracy : 0.7428 train accuracy : 0.7878 1395.29s epoch[ 6] test accuracy : 0.7513 train accuracy : 0.8045 1638.85s epoch[ 7] test accuracy : 0.7573 train accuracy : 0.8135 1882.32s epoch[ 8] test accuracy : 0.7638 train accuracy : 0.8221 fitting end