-
Notifications
You must be signed in to change notification settings - Fork 0
/
reproduce_reported_results_FB15K-237.py
179 lines (134 loc) · 6.42 KB
/
reproduce_reported_results_FB15K-237.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
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
from collections import defaultdict
import numpy as np
import torch
from models import ConExWithNorm
from helper_classes import Data
kg_path = 'KGs/FB15k-237'
data_dir = "%s/" % kg_path
model_path = 'PretrainedModels/FB15k-237/conex_fb_15k_237.pt'
d = Data(data_dir=data_dir, reverse=False)
class Reproduce:
def __init__(self):
self.cuda = False
self.batch_size = 256
def get_data_idxs(self, data):
data_idxs = [(self.entity_idxs[data[i][0]], self.relation_idxs[data[i][1]], self.entity_idxs[data[i][2]]) for i
in range(len(data))]
return data_idxs
def get_er_vocab(self, data):
er_vocab = defaultdict(list)
for triple in data:
er_vocab[(triple[0], triple[1])].append(triple[2])
return er_vocab
def get_batch(self, er_vocab, er_vocab_pairs, idx):
batch = er_vocab_pairs[idx:idx + self.batch_size]
targets = np.zeros((len(batch), len(d.entities)))
for idx, pair in enumerate(batch):
targets[idx, er_vocab[pair]] = 1.
targets = torch.FloatTensor(targets)
if self.cuda:
targets = targets.cuda()
return np.array(batch), targets
def evaluate(self, model, data, top_10_per_rel=True):
hits = []
ranks = []
rank_per_relation = dict()
for i in range(10):
hits.append([])
test_data_idxs = self.get_data_idxs(data)
inverse_relation_idx = dict(zip(self.relation_idxs.values(), self.relation_idxs.keys()))
er_vocab = self.get_er_vocab(self.get_data_idxs(d.data))
print("Number of data points: %d" % len(test_data_idxs))
for i in range(0, len(test_data_idxs), self.batch_size):
data_batch, _ = self.get_batch(er_vocab, test_data_idxs, i)
e1_idx = torch.tensor(data_batch[:, 0])
r_idx = torch.tensor(data_batch[:, 1])
e2_idx = torch.tensor(data_batch[:, 2])
if self.cuda:
e1_idx = e1_idx.cuda()
r_idx = r_idx.cuda()
e2_idx = e2_idx.cuda()
predictions = model.forward(e1_idx, r_idx)
for j in range(data_batch.shape[0]):
filt = er_vocab[(data_batch[j][0], data_batch[j][1])]
target_value = predictions[j, e2_idx[j]].item()
predictions[j, filt] = 0.0
predictions[j, e2_idx[j]] = target_value
sort_values, sort_idxs = torch.sort(predictions, dim=1, descending=True)
sort_idxs = sort_idxs.cpu().numpy()
for j in range(data_batch.shape[0]):
rank = np.where(sort_idxs[j] == e2_idx[j].item())[0][0]
ranks.append(rank + 1)
rank_per_relation.setdefault(inverse_relation_idx[data_batch[j][1]], []).append(rank + 1)
for hits_level in range(10):
val = 0.0
if rank <= hits_level:
val = 1.0
hits[hits_level].append(val)
print('Hits @10: {0}'.format(np.mean(hits[9])))
print('Hits @3: {0}'.format(np.mean(hits[2])))
print('Hits @1: {0}'.format(np.mean(hits[0])))
print('Mean rank: {0}'.format(np.mean(ranks)))
print('Mean reciprocal rank: {0}'.format(np.mean(1. / np.array(ranks))))
print('##########################################################\n')
if top_10_per_rel:
# Get frequencies
freq = dict()
for i in data:
rel = i[1]
if rel in freq:
freq[rel] += 1
else:
freq[rel] = 1
freq = {k: v for k, v in sorted(freq.items(), key=lambda item: item[1], reverse=True)}
selected_rels = []
# Select top 10
top = 1
for rel, v in freq.items():
print('{0}. {1} => freq {2}'.format(top, rel, v))
selected_rels.append(rel)
if top == 10:
break
top += 1
for rel in selected_rels:
# data conversion
rank_per_relation[rel] = np.array(rank_per_relation[rel])
print('{0}: Mean Reciprocal Rank: {1}'.format(rel, np.mean(1. / rank_per_relation[rel])))
hit10 = (rank_per_relation[rel] <= 10).sum() / len(rank_per_relation[rel])
hit3 = (rank_per_relation[rel] <= 3).sum() / len(rank_per_relation[rel])
hit1 = (rank_per_relation[rel] == 1).sum() / len(rank_per_relation[rel])
print('H@10 for {0}: {1}'.format(rel, hit10))
print('H@3 for {0}: {1}'.format(rel, hit3))
print('H@1 for {0}: {1}'.format(rel, hit1))
else:
for relations, ranks_ in rank_per_relation.items():
rank_per_relation[relations] = np.array(ranks_)
print('{0}: Mean Reciprocal Rank: {1}'.format(relations, np.mean(1. / rank_per_relation[relations])))
hit10 = (rank_per_relation[relations] <= 10).sum() / len(rank_per_relation[relations])
hit3 = (rank_per_relation[relations] <= 3).sum() / len(rank_per_relation[relations])
hit1 = (rank_per_relation[relations] == 1).sum() / len(rank_per_relation[relations])
print('H@10 for {0}: {1}'.format(relations, hit10))
print('H@3 for {0}: {1}'.format(relations, hit3))
print('H@1 for {0}: {1}'.format(relations, hit1))
def reproduce(self):
self.entity_idxs = {d.entities[i]: i for i in range(len(d.entities))}
self.relation_idxs = {d.relations[i]: i for i in range(len(d.relations))}
params = {'num_entities': len(self.entity_idxs),
'num_relations': len(self.relation_idxs),
'embedding_dim': 20,
'input_dropout': 0.1,
'hidden_dropout': 0.1,
'projection_size': 1088,
'conv_out': 64,
'feature_map_dropout': 0.1}
model = ConExWithNorm(params)
model.load_state_dict(torch.load(model_path))
for parameter in model.parameters():
parameter.requires_grad = False
model.eval()
if self.cuda:
model.cuda()
print('Number of free parameters: ', sum([p.numel() for p in model.parameters()]))
print('Test Results')
self.evaluate(model, d.test_data)
Reproduce().reproduce()