default=1500)
(options, args) = parser.parse_args()
########
trainSize = options.trainSize
resultDir = options.resultDir
modelName = options.modelName
trainFlag = options.trainFlag
devFlag = options.devFlag
testFlag = options.testFlag
numPowerTraces = options.numPowerTraces
dataDir = "/xdisk/rlysecky/manojgopale/extra/chipWhisperer_data/trace_key_1500/"
moreDataDir = "/xdisk/rlysecky/manojgopale/extra/chipWhisperer_data/trace_key_1500_1/"
trainData, devData, testData = classify_general.getData(
dataDir, moreDataDir, trainSize, trainFlag, devFlag, testFlag)
x_train, y_train_oh = trainData
x_dev, y_dev_oh = devData
x_test, y_test_oh = testData
## Instantiate the model and test, dev and training sets
##resultDir = "/extra/manojgopale/AES_data/config3p1_15ktraining/result_new"
##modelName = "m_newscript"
np.random.seed()
numHiddenLayers = 4
actList = ['elu', 'tanh', 'relu', 'relu']
dropList = [0.3, 0.2, 0.5, 0.25]
batchNorm = [1, 0, 0, 1]
batchSize = 16000
default=1500)
(options, args) = parser.parse_args()
########
trainSize = options.trainSize
resultDir = options.resultDir
modelName = options.modelName
configName = options.configName
trainFlag = options.trainFlag
devFlag = options.devFlag
testFlag = options.testFlag
numPowerTraces = options.numPowerTraces
dataDir = "/xdisk/rlysecky/manojgopale/xdisk/gem5DataCollection/csvResult/"
trainData, devData, testData = classify_general.getData(
dataDir, configName, trainSize, trainFlag, devFlag, testFlag)
x_train, y_train_oh = trainData
x_dev, y_dev_oh = devData
x_test, y_test_oh = testData
## Instantiate the model and test, dev and training sets
##resultDir = "/extra/manojgopale/AES_data/config3p1_15ktraining/result_new"
##modelName = "m_newscript"
np.random.seed()
numAllHiddenLayers = [3, 4, 5, 6, 7, 8, 9, 10]
hiddenLayerDict = {
"num": [1, 2, 3, 4, 5, 6, 7, 8, 9, 10],
"factor": [10, 100, 1000]
}
modelDir = options.modelDir
devDataDir = options.devDataDir
trainFlag = options.trainFlag
devFlag = options.devFlag
testFlag = options.testFlag
count = options.count
########
### TODO:
### 1. Load dev data
### 2. Load models one-by-one and evaluate
### 3. Create csv/tsv files and run error_analysis on it
devDir = devDataDir + devDataConfig + "/"
_, devData, _ = classify_general.getData(devDataDir, devDataConfig, trainSize, trainFlag, devFlag, testFlag)
x_dev, y_dev_oh = devData
modelDict = OrderedDict()
modelDict = {\
"config3p1": "rerun_cw_run_177_4HL_11epochs_100p00_acc_.h5",\
"config3p2": "rerun_cw_run_config3p2_177_4HL_11epochs_98p05_acc_.h5",\
"config3p3": "rerun_cw_run_177_4HL_11epochs_98p05_acc_.h5",\
"config3p4": "rerun_cw_run_config3p4_177_4HL_17epochs_100p00_acc_.h5",\
"config4p1": "rerun_cw_run_177_4HL_11epochs_97p66_acc_.h5",\
"config4p2": "rerun_cw_run_177_4HL_11epochs_97p66_acc_.h5",\
"config4p3": "rerun_cw_run_config4p3_177_4HL_17epochs_100p00_acc_.h5",\
"config4p4": "rerun_cw_run_177_4HL_11epochs_98p05_acc_.h5",\
"config5p1": "rerun_cw_run_177_4HL_18epochs_100p00_acc_.h5",\
"config5p2": "rerun_cw_run_config5p2_177_4HL_18epochs_100p00_acc_.h5",\
"config5p3": "rerun_cw_run_177_4HL_11epochs_97p66_acc_.h5",\
"config3p1_1_4HL_8epochs_97p86_acc_.h5",
"config3p2_1_4HL_7epochs_99p99_acc_.h5",
"config3p3_1_4HL_6epochs_99p98_acc_.h5",
"config3p4_1_4HL_7epochs_99p97_acc_.h5",
"config4p1_1_4HL_12epochs_99p99_acc_.h5",
"config4p2_1_4HL_11epochs_100p00_acc_.h5",
"config4p3_1_4HL_7epochs_99p99_acc_.h5",
"config4p4_1_4HL_8epochs_99p94_acc_.h5",
"config5p1_17_4HL_6epochs_100p00_acc_.h5",
"config5p2_17_4HL_7epochs_98p38_acc_.h5",
"config5p3_17_4HL_6epochs_100p00_acc_.h5",
"config5p4_17_4HL_13epochs_70p49_acc_.h5"
]
if (devFlag):
_, devData, _ = classify_general.getData(testDir, testConfig, trainSize,
trainFlag, devFlag, testFlag)
x_dev, y_dev_oh = devData
for index, modelName in enumerate(modelList):
modelPath = modelDir + "/" + configList[count] + "/" + modelList[count]
print("Loading model \n%s\n" % (modelPath))
model = load_model(modelPath)
print("Finished loading model \n%s\n" % (modelPath))
## Evaluate the performance of model on testData
model_score = model.evaluate(x_dev, y_dev_oh, batch_size=2048)
print("\nmodel= %s score on %s is: %s\n" %
(configList[count], testConfig, model_score[1]))
## Convert from one-hot to numerical prediction
y_pred = np.argmax(model.predict(x_dev, batch_size=2048), axis=1)