-
Notifications
You must be signed in to change notification settings - Fork 4
/
Copy pathfusion.py
174 lines (134 loc) · 6.79 KB
/
fusion.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
import os
from kge_model import KGEModel
from dataloader import *
import torch
import torch.nn as nn
import torch.optim as optim
import numpy as np
import logging
from tqdm import tqdm
def train_fusion(args, all_data, num_clients, fusion_state):
one_client_state_str, fed_state_str = fusion_state
fed_state_str = f'{fed_state_str}.best'
result_list = []
test_len_list = np.zeros(num_clients)
for i in range(num_clients):
data = all_data[i]
curr_client_state_str = f'{one_client_state_str}_client_{i}.best'
res, test_len = fusion_on_client(args, i, data, curr_client_state_str, fed_state_str)
result_list.append(res)
test_len_list[i] = test_len
test_len_list = test_len_list / np.sum(test_len_list)
results = ddict(int)
for i in range(num_clients):
for k, v in result_list[i].items():
results[k] += test_len_list[i] * v
logging.info('overall result:')
logging.info('mrr: {:.4f}, hits@1: {:.4f}, hits@5: {:.4f}, hits@10: {:.4f}'.format(
results['mrr'], results['hits@1'],
results['hits@5'], results['hits@10']))
def fusion_on_client(args, idx, data, one_client_state_str, fed_state_str):
kge_model = KGEModel(args, model_name=args.model)
# trained embedding
state = torch.load(os.path.join(args.state_dir, one_client_state_str),
map_location=args.gpu)
rel_embed = state['rel_emb'].detach()
ent_embed = state['ent_emb'].detach()
fed_state = torch.load(os.path.join(args.state_dir, fed_state_str),
map_location=args.gpu)
rel_embed_fed = fed_state['rel_embed'][idx].detach()
ent_embed_fed = fed_state['ent_embed'].detach()
nentity = len(np.unique(data['train']['edge_index'].reshape(-1)))
ent_purm = np.zeros(nentity, dtype=np.int64)
for i in range(data['train']['edge_index'].shape[1]):
h, r, t = data['train']['edge_index'][0][i], data['train']['edge_type'][i], \
data['train']['edge_index'][1][i]
h_ori, r_ori, t_ori = data['train']['edge_index_ori'][0][i], data['train']['edge_type_ori'][i], \
data['train']['edge_index_ori'][1][i]
ent_purm[h] = h_ori
ent_purm[t] = t_ori
ent_purm = torch.LongTensor(ent_purm)
ent_embed_fed = ent_embed_fed[ent_purm]
# dataloader
nentity = len(np.unique(data['train']['edge_index'].reshape(-1)))
nrelation = len(np.unique(data['train']['edge_type']))
train_triples = np.stack((data['train']['edge_index'][0],
data['train']['edge_type'],
data['train']['edge_index'][1])).T
valid_triples = np.stack((data['valid']['edge_index'][0],
data['valid']['edge_type'],
data['valid']['edge_index'][1])).T
test_triples = np.stack((data['test']['edge_index'][0],
data['test']['edge_type'],
data['test']['edge_index'][1])).T
all_triples = np.concatenate([train_triples, valid_triples, test_triples])
# valid_train_dataset = TrainDataset(valid_triples, nentity, args.num_neg, 'tail-batch', all_triples)
valid_train_dataset = TrainDataset(valid_triples, nentity, args.num_neg)
test_dataset = TestDataset(test_triples, all_triples, nentity, 'tail-batch')
valid_train_dataloader = DataLoader(
valid_train_dataset,
batch_size=args.batch_size,
collate_fn=TrainDataset.collate_fn
)
test_dataloader = DataLoader(
test_dataset,
batch_size=args.test_batch_size,
collate_fn=TestDataset.collate_fn
)
# linear model
linear = nn.Linear(in_features=2, out_features=1).to(args.gpu)
criterion = nn.MarginRankingLoss(10, reduction='mean')
optimizer = optim.Adam(linear.parameters(), lr=0.01)
t = tqdm(range(500))
for epoch in t:
losses = []
for batch in valid_train_dataloader:
positive_sample, negative_sample, _ = batch
positive_sample = positive_sample.to(args.gpu)
negative_sample = negative_sample.to(args.gpu)
negative_score = kge_model((positive_sample, negative_sample),
rel_embed, ent_embed)
negative_score_fed = kge_model((positive_sample, negative_sample),
rel_embed_fed, ent_embed_fed)
positive_score = kge_model(positive_sample, rel_embed, ent_embed, neg=False).squeeze(dim=1)
positive_score_fed = kge_model(positive_sample, rel_embed_fed, ent_embed_fed, neg=False).squeeze(dim=1)
neg_score = torch.cat([negative_score.unsqueeze(2), negative_score_fed.unsqueeze(2)], dim=-1)
pos_score = torch.cat([positive_score.unsqueeze(1), positive_score_fed.unsqueeze(1)], dim=-1)
neg_out = linear(neg_score)
pos_out = linear(pos_score)
loss = criterion(pos_out, torch.mean(neg_out, dim=-1), torch.LongTensor([1]).to(args.gpu))
t.set_postfix({'loss': '{:.4f}'.format(loss)})
optimizer.zero_grad()
loss.backward()
optimizer.step()
losses.append(loss.item())
torch.save(linear.state_dict(), os.path.join(args.state_dir, one_client_state_str + '.fusion'))
results = ddict(float)
for batch in test_dataloader:
triplets, labels = batch
triplets, labels = triplets.to(args.gpu), labels.to(args.gpu)
head_idx, rel_idx, tail_idx = triplets[:, 0], triplets[:, 1], triplets[:, 2]
pred = kge_model((triplets, None), rel_embed, ent_embed)
pred_multi = kge_model((triplets, None), rel_embed_fed, ent_embed_fed)
pred = torch.cat([pred.unsqueeze(-1), pred_multi.unsqueeze(-1)], dim=-1)
pred = linear(pred).squeeze(-1)
b_range = torch.arange(pred.size()[0], device=args.gpu)
target_pred = pred[b_range, tail_idx]
pred = torch.where(labels.byte(), -torch.ones_like(pred) * 10000000, pred)
pred[b_range, tail_idx] = target_pred
pred_argsort = torch.argsort(pred, dim=1, descending=True)
ranks = 1 + torch.argsort(pred_argsort, dim=1, descending=False)[b_range, tail_idx]
ranks = ranks.float()
count = torch.numel(ranks)
results['count'] += count
results['mr'] += torch.sum(ranks).item()
results['mrr'] += torch.sum(1.0 / ranks).item()
for k in [1, 5, 10]:
results['hits@{}'.format(k)] += torch.numel(ranks[ranks <= k])
for k, v in results.items():
if k != 'count':
results[k] /= results['count']
logging.info('mrr: {:.4f}, hits@1: {:.4f}, hits@5: {:.4f}, hits@10: {:.4f}'.format(
results['mrr'], results['hits@1'],
results['hits@5'], results['hits@10']))
return results, len(test_dataloader.dataset)