return model
if __name__ == "__main__":
"""
train test GRU:
---
Test loss: 0.6381104083000859
Test accuracy: 0.7306
"""
num_classes = 3
epochs = 3
vocab_set = load_pickle('./preprocess/vocab_set.pkl')
# vocab_set = {}
train_data, test_data, params, vocab_set = get_test_dataset(
epochs=epochs, vocab_set=vocab_set, max_len=30, cache_dir='cache/base')
max_len = params.get('max_len')
num_words = params.get('vocab_size')
batch_size = params.get('batch_size')
model = build_gru_model(max_len, num_words, num_classes)
model.compile(loss='sparse_categorical_crossentropy',
optimizer='adam',
metrics=['accuracy'])
model.summary()
model.fit(train_data,
epochs=epochs,
validation_data=test_data)
score = model.evaluate(test_data, verbose=0)
if __name__ == "__main__":
'''
Train test char-level convolution:
---
Test loss: 0.6190159400052662
Test accuracy: 0.6076
'''
# const
num_classes = 3
epochs = 3
vocab_set = load_pickle('./preprocess/vocab_set_char.pkl')
# vocab_set = {}
train_data, test_data, params, vocab_set = get_test_dataset(
epochs=epochs,
vocab_set=vocab_set,
max_len=62,
char=True,
cache_dir='cache/char')
# save_pickle(vocab_set, './preprocess/vocab_set_char.pkl')
hyperparams = {
'maxlen': params.get('max_len'),
'num_words': params.get('vocab_size'),
'num_classes': num_classes,
'activation': "softmax",
'mingram': 2,
'maxgram': 5,
'emb_dim': 32,
'filter_size': 32,
'num_dense': 2,