search

DwangoMediaVillage / keras_compressor

Model Compression CLI Tool for Keras.
https://nico-opendata.jp/ja/casestudy/model_compression/index.html
157 stars 39 forks source link

Not compressed #13

Open Bashima opened 5 years ago

Bashima commented 5 years ago

There is no compression. I have a the following layers conv1D>batch_norm>activation> max_pool

Command: keras-compressor.py --error 0.1 with.h5 com.h5

OUTPUT Using TensorFlow backend. 2018-10-16 13:09:04.704740: I tensorflow/core/platform/cpu_feature_guard.cc:141] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 FMA INFO:keras_compressor.compressor:factorizer not found layer:<keras.layers.convolutional.Conv1D object at 0x7fa81f200080> INFO:keras_compressor.compressor:factorizer not found layer:<keras.layers.normalization.BatchNormalization object at 0x7fa81f1eab00> INFO:keras_compressor.compressor:factorizer not found layer:<keras.layers.core.Activation object at 0x7fa81f1eab38> INFO:keras_compressor.compressor:factorizer not found layer:<keras.layers.pooling.MaxPooling1D object at 0x7fa81f1ae710> INFO:keras_compressor.compressor:factorizer not found layer:<keras.layers.convolutional.Conv1D object at 0x7fa81f1ea898> INFO:keras_compressor.compressor:factorizer not found layer:<keras.layers.normalization.BatchNormalization object at 0x7fa81f0e9080> INFO:keras_compressor.compressor:factorizer not found layer:<keras.layers.core.Activation object at 0x7fa81f1a2e48> INFO:keras_compressor.compressor:factorizer not found layer:<keras.layers.pooling.MaxPooling1D object at 0x7fa81f11e048> INFO:keras_compressor.compressor:factorizer not found layer:<keras.layers.convolutional.Conv1D object at 0x7fa81f0e9ef0> INFO:keras_compressor.compressor:factorizer not found layer:<keras.layers.normalization.BatchNormalization object at 0x7fa81f0381d0> INFO:keras_compressor.compressor:factorizer not found layer:<keras.layers.core.Activation object at 0x7fa81f085ef0> INFO:keras_compressor.compressor:factorizer not found layer:<keras.layers.pooling.MaxPooling1D object at 0x7fa81edcb358> INFO:keras_compressor.compressor:factorizer not found layer:<keras.layers.convolutional.Conv1D object at 0x7fa81edddbe0> INFO:keras_compressor.compressor:factorizer not found layer:<keras.layers.normalization.BatchNormalization object at 0x7fa81ed5ee10> INFO:keras_compressor.compressor:factorizer not found layer:<keras.layers.core.Activation object at 0x7fa81ed2a710> INFO:keras_compressor.compressor:factorizer not found layer:<keras.layers.pooling.MaxPooling1D object at 0x7fa81f038b00> INFO:keras_compressor.compressor:factorizer not found layer:<keras.layers.core.Lambda object at 0x7fa81ed2ae48> INFO:keras_compressor.compressor:factorizer found layer:<keras.layers.core.Dense object at 0x7fa81ed1e860> factorizer:<class 'keras_compressor.factorizers.svd.SVDFactorizer'> INFO:keras_compressor.compressor:factorization failed layer:<keras.layers.core.Dense object at 0x7fa81ed1e860> factorizer:<class 'keras_compressor.factorizers.svd.SVDFactorizer'> Compressed model before #params 561,162 after #params 561,162 (0.00%)

suzy8808 commented 5 years ago

My model isn't compressed too

EricHuiK commented 4 years ago

My model isn't compressed too （keras-retinaNet）