from model.lstm_model import LSTMModel
from model.persistence.model_persistence import ModelEvaluation
from utils.nn_utils import NNUtils
from utils.config_loader import readConfig
def showProgress(currentStep, totalSteps, epoch):
perc = (float(currentStep) / float(totalSteps)) * 100.0
temp = perc / 10
sys.stdout.write('\r[{0}] {1}% - {2}/{3} - Epoch {4}'.format(
'#' * int(temp), (perc), currentStep, totalSteps, epoch))
sys.stdout.flush()
print 'loading configuration'
config, modelConfig = readConfig()
print 'configuration loaded'
print 'loading word embeddings : {} - embedding size : {}'.format(
modelConfig.embeddingType, modelConfig.embeddingSize)
sentenceLoader, predicateLoader = getEmbeddings()
print 'sentenceLoader shape {}'.format(sentenceLoader.weights.shape)
nnUtils = NNUtils.Instance()
nnUtils.setWordUtils(sentenceLoader.word2idx, sentenceLoader.idx2word)
print 'loaded'
print 'loading corpus'
csvFiles = [
config.convertedCorpusDir + '/propbank_training.csv',
sentences, predicates, aux, roles)
# the order is important
structure.append((sentences, predicates, aux, roles))
return structure
def convertAndSave(self, featureFile):
temp = self.convert()
self.save(temp, featureFile)
if __name__ == '__main__':
from embeddings.emb_loader import W2VModel
from utils.config_loader import readConfig
readConfig()
options = {
"npzFile": "../../resources/embeddings/wordEmbeddings.npy",
"npzModel": "../../resources/embeddings/wordEmbeddings",
"vecFile": "../../resources/embeddings/model.vec",
"w2idxFile": "../../resources/embeddings/vocabulary.json"
}
model = W2VModel()
model.setResources(options)
loader = EmbeddingLoader(model)
loader.process()
csvFiles = [
'../../resources/corpus/converted/propbank_training.csv',
'../../resources/corpus/converted/propbank_test.csv'
]