-
Notifications
You must be signed in to change notification settings - Fork 10
/
train_medium.py
44 lines (44 loc) · 1.86 KB
/
train_medium.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
import pickle
import argparse
import sys
import numpy as np
parser = argparse.ArgumentParser()
parser.add_argument("-m", "--model", help="the model chosen", default="internal", type=str)
args = parser.parse_args()
model_chosen = args.model
if model_chosen == "internal":
from internal_mem import model_revise
from internal_mem import solver_revise
ECM_Model = model_revise.ECM_Model
ECMSolver = solver_revise.ECMSolver
else:
from emo_embedding import model_revise
from emo_embedding import solver_revise
ECM_Model = model_revise.ECM_Model
ECMSolver = solver_revise.ECMSolver
f = open('data_medium.pkl', 'rb')
data = pickle.load(f)
f.close()
f = open('word2idx_medium.pkl', 'rb')
word2idx = pickle.load(f)
max_value = max(word2idx.values())
f.close()
max_length_questions = 0
max_length = 0
for question in data['questions']:
max_length_questions = max(max_length_questions, len(question))
for answer in data['answers']:
max_length = max(max_length, len(answer))
for question in data['trans_questions']:
question.extend([word2idx['<NULL>']] * (max_length_questions-len(question)))
for answer in data['trans_answers']:
answer.extend([word2idx['<NULL>']] * (max_length-len(answer)))
batch_size = 128
n_iters_per_epoch = int(np.ceil(float(len(data['questions'])) / batch_size))
model = ECM_Model(max_length_questions, max_length, emotion_num=6, word_to_idx=word2idx, embedding_matrix=None, learning_rate=0.5)
solver = ECMSolver(model, data, word2idx=word2idx, val_data=None, n_epochs=2000, batch_size=batch_size, update_rule='adam',
print_every=int(n_iters_per_epoch/2), save_every=10,
pretrained_model=None, model_path='model/lstm/'+model_chosen,
test_model='model/lstm/model-10',
log_path='log/')
solver.train()