def main(opts):
RECEIVED_PARAMS = nni.get_next_parameter()
LOG.debug(RECEIVED_PARAMS)
if 'onlyIncoming' == opts.mode:
PARAMS = cudnnLstm.generate_default_onlyIncoming_params(opts.dataType)
elif 'both' == opts.mode:
PARAMS = cudnnLstm.generate_default_whole_params(opts.dataType)
PARAMS.update(RECEIVED_PARAMS)
try:
X_train_raw, y_train, X_test_raw, y_test, labelMap = loadTrainAndTestData(
opts.input, PARAMS['data_dim'], opts.dataType, opts.mode)
NUM_CLASS = len(set(y_test))
X_train = X_train_raw.reshape(X_train_raw.shape[0],
X_train_raw.shape[1], 1)
X_test = X_test_raw.reshape(X_test_raw.shape[0], X_test_raw.shape[1],
1)
y_train = np_utils.to_categorical(y_train, NUM_CLASS)
y_test = np_utils.to_categorical(y_test, NUM_CLASS)
lstm_model = LSTM(opts, PARAMS)
modelPath = lstm_model.train(PARAMS, X_train, y_train, NUM_CLASS)
lstm_model.test(X_test, y_test, NUM_CLASS, modelPath)
except Exception as e:
LOG.exception(e)
raise
def loadData(opts, PARAMS):
X_train, y_train, X_test, y_test, labelMap = prepareData.loadTrainAndTestData(
opts.input, PARAMS['data_dim'], opts.dataType, opts.mode)
NUM_CLASS = len(set(y_test))
y_train = np_utils.to_categorical(y_train, NUM_CLASS)
y_test = np_utils.to_categorical(y_test, NUM_CLASS)
return X_train, y_train, X_test, y_test, labelMap, NUM_CLASS
def loadData(opts, params):
X_train_raw, y_train, X_test_raw, y_test, labelMap = loadTrainAndTestData(
opts.input, params['data_dim'], opts.dataType)
NUM_CLASS = len(set(y_test))
X_train = X_train_raw.reshape(X_train_raw.shape[0], X_train_raw.shape[1],
1)
X_test = X_test_raw.reshape(X_test_raw.shape[0], X_test_raw.shape[1], 1)
y_train = np_utils.to_categorical(y_train, NUM_CLASS)
y_test = np_utils.to_categorical(y_test, NUM_CLASS)
return X_train, y_train, X_test, y_test, labelMap, NUM_CLASS
def main(opts):
RECEIVED_PARAMS = nni.get_next_parameter()
LOG.debug(RECEIVED_PARAMS)
if 'onlyIncoming' == opts.mode:
PARAMS = sae.generate_default_onlyIncoming_params(opts.dataType)
elif 'both' == opts.mode:
PARAMS = sae.generate_default_whole_params(opts.dataType)
PARAMS.update(RECEIVED_PARAMS)
try:
X_train, y_train, X_test, y_test, labelMap = prepareData.loadTrainAndTestData(opts.input, PARAMS['data_dim'], opts.dataType, opts.mode)
NUM_CLASS = len(set(y_test))
y_train = np_utils.to_categorical(y_train, NUM_CLASS)
y_test = np_utils.to_categorical(y_test, NUM_CLASS)
sae_model = SAE(opts, PARAMS)
modelPath = sae_model.train(PARAMS, X_train, y_train, NUM_CLASS)
sae_model.test(X_test, y_test, NUM_CLASS, modelPath)
except Exception as e:
LOG.exception(e)
raise