Define data paths to use in a config.ini file:
[data_path] BASE_DIR=../NLP WORD_EMBED=%(BASE_DIR)s/word_embed/glove.6B.300d.txt SNLI=%(BASE_DIR)s/SNLI/snli_1.0 TRAIN_DATA=%(SNLI)s/snli_1.0_train.txt DEV_DATA=%(SNLI)s/snli_1.0_dev.txt TEST_DATA=%(SNLI)s/snli_1.0_test.txt
You can generate the data for input as follows. Once saving the generated data, you can quickly load the data:
from SNLI import SNLI
snli = SNLI()
# generate data
snli.generate_data('../data/snli_concat_glove6B.300d.pickle')
# load data
snli = snli.load_data('../data/snli_concat_glove6B.300d.pickle')