ML-KWS-for-MCU
ML-KWS-for-MCU copied to clipboard
ARM-software
Error run quant_test.py to quantize ds_cnn
INFO:tensorflow:Confusion Matrix: [[ 0 3124 0 0 0 0 0 0 0 0 0 0] [ 0 3126 0 0 0 0 0 0 0 0 0 0] [ 0 3289 0 0 0 0 0 0 0 0 0 0] [ 0 3209 0 0 0 0 0 0 0 0 0 0] [ 0 3038 0 0 0 0 0 0 0 0 0 0] [ 0 3009 0 0 0 0 0 0 0 0 0 0] [ 0 3026 0 0 0 0 0 0 0 0 0 0] [ 0 3004 0 0 0 0 0 0 0 0 0 0] [ 0 3004 0 0 0 0 0 0 0 0 0 0] [ 0 2976 0 0 0 0 0 0 0 0 0 0] [ 0 3078 0 0 0 0 0 0 0 0 0 0] [ 0 3040 0 0 0 0 0 0 0 0 0 0]] INFO:tensorflow:Training accuracy = 8.47% (N=36923) INFO:tensorflow:set_size=4445 INFO:tensorflow:Confusion Matrix: [[ 0 371 0 0 0 0 0 0 0 0 0 0] [ 0 371 0 0 0 0 0 0 0 0 0 0] [ 0 397 0 0 0 0 0 0 0 0 0 0] [ 0 406 0 0 0 0 0 0 0 0 0 0] [ 0 350 0 0 0 0 0 0 0 0 0 0] [ 0 377 0 0 0 0 0 0 0 0 0 0] [ 0 352 0 0 0 0 0 0 0 0 0 0] [ 0 363 0 0 0 0 0 0 0 0 0 0] [ 0 363 0 0 0 0 0 0 0 0 0 0] [ 0 373 0 0 0 0 0 0 0 0 0 0] [ 0 350 0 0 0 0 0 0 0 0 0 0] [ 0 372 0 0 0 0 0 0 0 0 0 0]] INFO:tensorflow:Validation accuracy = 8.35% (N=4445) INFO:tensorflow:set_size=4890 INFO:tensorflow:Confusion Matrix: [[ 0 408 0 0 0 0 0 0 0 0 0 0] [ 0 408 0 0 0 0 0 0 0 0 0 0] [ 0 419 0 0 0 0 0 0 0 0 0 0] [ 0 405 0 0 0 0 0 0 0 0 0 0] [ 0 425 0 0 0 0 0 0 0 0 0 0] [ 0 406 0 0 0 0 0 0 0 0 0 0] [ 0 412 0 0 0 0 0 0 0 0 0 0] [ 0 396 0 0 0 0 0 0 0 0 0 0] [ 0 396 0 0 0 0 0 0 0 0 0 0] [ 0 402 0 0 0 0 0 0 0 0 0 0] [ 0 411 0 0 0 0 0 0 0 0 0 0] [ 0 402 0 0 0 0 0 0 0 0 0 0]] INFO:tensorflow:Test accuracy = 8.34% (N=4890)
Why? When I changed the set of hyperparameters (window_size, window_stride, etc.), it still did not work. However, there is a right result when quantizing dnn model.
run fold_batchnorm.py, after use "-bnfused" checkpoint to run quant_test.py check "-act_max []" argument in quant_test.py
Don't change any hyper-parameters, training and testing parameters must be same
run fold_batchnorm.py, after use "-bnfused" checkpoint to run quant_test.py check "-act_max []" argument in quant_test.py
Don't change any hyper-parameters, training and testing parameters must be same
Thank you a lot! I got it!
I used the fold_batchnorm.py, and after creating the "-bnfused" checkpoint,I ran quant_test.py with it. The accuracy did not change comparing to the time I used the simple checkpoint.(For DNN the accuracy for both was about 84%).Do you know if this is supposed to be like that? or is this wrong?
I used the fold_batchnorm.py, and after creating the "-bnfused" checkpoint,I ran quant_test.py with it. The accuracy did not change comparing to the time I used the simple checkpoint.(For DNN the accuracy for both was about 84%).Do you know if this is supposed to be like that? or is this wrong?
There are not batchnorm layers in DNN model, so it's not meaningful to do that with DNN model.