r_dict[r[i]] = count
count += 1
if __name__ == "__main__":
parser = argparse.ArgumentParser()
parser.add_argument("--repository_path", type=str, required=True)
parser.add_argument("--data_path", type=str, required=True)
parser.add_argument("--output_path", type=str, required=True)
args = parser.parse_args()
repository_path = args.repository_path
data_path = args.data_path
tr, v, tst = read_data(data_path)
ratios = utils.read_qprobs(repository_path)
create_ratio_dict(ratios)
print("Processing training set")
tr_inp, tr_m_out, tr_q_out, tr_r_out, tr_inp_names, tr_inp_years = load_data(
tr)
print("Processing validation set")
v_inp, v_m_out, v_q_out, v_r_out, v_inp_names, v_inp_years = load_data(v)
print("Processing test set")
t_inp, t_m_out, t_q_out, t_r_out, t_inp_names, t_inp_years = load_data(tst)
token2id = {}
id2token = {}
valid = pickle.load(in_file)
dataset_v = valid["dataset_v"]
v_m_out = valid["v_m_out"]
v_q_out = valid["v_q_out"]
v_r_out = valid["v_r_out"]
print("Loading filename: {}".format(args.embeddings_filename))
embeddings_index = {}
with open(args.embeddings_filename) as in_file:
for line in in_file:
values = line.split()
word = values[0]
coefs = np.asarray(values[1:], dtype='float32')
embeddings_index[word] = coefs
embedding_matrix = np.zeros((len(token2id) + 1, 300))
for word, i in token2id.items():
embedding_vector = embeddings_index.get(word)
if embedding_vector is not None:
embedding_matrix[i] = embedding_vector
# Load visual data
tr_inds, v_inds, t_inds = read_indices(args.vision_dataset_path)
ratios = utils.read_qprobs(args.vision_dataset_path)
tr_size = 11900
v_size = 1700
create_ratio_dict(ratios)
tr_inp, tr_m_out, tr_q_out, tr_r_out = read_images(tr_inds, tr_size)
v_inp, v_m_out, v_q_out, v_r_out = read_images(v_inds, v_size)