-
Notifications
You must be signed in to change notification settings - Fork 0
/
CKRL.py
35 lines (26 loc) · 903 Bytes
/
CKRL.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
import os
import numpy as np
import re
import pickle as pk
import random
from numpy import linalg as LA
from CorrelationEmbedding import *
from InputDataStructure import *
from DataPreproccessing import *
from DataModel import *
from ModelEvaluator import *
train_path = r'Dataset\FB15K-237\train.txt'
test_path = r'Dataset\FB15K-237\test.txt'
save_input_data_path = r'Data\dataset'
save_preproccessing_data_path = r'Data\preproccessed_data'
save_model_path = r'Data\data_model.dat'
evaluation_result_path = r'Data\evaluation.dat'
CE = CorrelationEmbedding()
embedding_size = 50
CE.load_datesets(train_path, test_path, save_input_data_path)
CE.preproccess(save_preproccessing_data_path, embedding_size)
model = CE.data_model(save_model_path, embedding_size)
max_iter = 1000
for cur_iter in range(max_iter):
print(f'{cur_iter}/{max_iter}')
model.train()