-
Notifications
You must be signed in to change notification settings - Fork 16
/
inference.py
166 lines (149 loc) · 6.13 KB
/
inference.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
#!user/bin/env python
# -*- coding:utf-8 -*-
from dataset import vocab_num, my_collate
from dataset_val import KgDatasetVal
from dataset import KgDataset
import torch
import argparse
from torch.utils.data import DataLoader
import torch.optim as optim
import torch.nn as nn
import os
import random
import torch.nn.functional as F
from tqdm import tqdm
import pickle
from transformers import LxmertTokenizer
from train import my_collate, cal_batch_loss, generate_tripleid, cal_acc_multi
from model import KgPreModel, tokenizer
from config import args
import json
import numpy as np
torch.multiprocessing.set_sharing_strategy('file_system')
device = torch.device("cuda") if torch.cuda.is_available() else torch.device("cpu")
def my_collate_2(batch):
id = []
ques = []
ans = []
img = []
object = []
spatial = []
length = []
for item in batch:
id.append(item['id'])
ques.append(item['ques'])
ans.append(item['ans'])
img.append(item['img'])
object.append(item['object'])
spatial.append(item['spatial'])
length.append(item['length'])
res = {'id': id, 'ques': ques, 'ans': ans,
'img': img, 'length': length,
'object': object, 'spatial': spatial}
return res
def write_down_prediction(object_list, qid, preds, ans, mean, length_list):
assert len(preds) == len(qid)
write_down_dict = {}
mean_dict = {}
s = 0
for i, answer_id in enumerate(qid):
predict_objects = []
pred = preds[i].squeeze()
length = length_list[i] + 36
pred = pred[:length]
_, idx_1 = torch.topk(pred, k=1)
predict_objects = object_list[i][idx_1]
# idx_3 = idx_3.tolist()
# for idx in idx_3:
# try:
# predict_objects.append(object_list[i][idx])
# except IndexError:
# print(pred)
# s += 1
write_down_dict[answer_id] = predict_objects#(predict_objects, ans[i])
mean_dict[answer_id] = mean[i].cpu().numpy()
# print(s)
with open("krisp_full_generate_v1_val.json", 'w') as f:
json.dump(write_down_dict, f, indent=4)
with open('mean_out_temp.pickle', 'wb') as f:
pickle.dump(mean_dict, f)
def test():
if args.embedding:
answer_candidate_tensor = torch.arange(0, vocab_num).view(-1, 1).long().cuda()
# else:
# answer_candidate_tensor = torch.tensor(answer_embedding).float().cuda()
# answer_candidate_tensor = F.normalize(answer_candidate_tensor, dim=1, p=2)
test_dataset = KgDatasetVal()
test_dataloader = DataLoader(dataset=test_dataset, batch_size=args.batch_size, shuffle=False,
num_workers=0, collate_fn=my_collate)
model = KgPreModel(vocab_num)
model = model.to(device)
model.eval()
for epoch in range(1, 4):
path = args.model_dir + 'model_for_epoch_%d.pth' % epoch
print(f"\nValidation after epoch {epoch}:")
model.load_state_dict(torch.load(path))
preds = []
ground_ts = []
qids = []
object_lists = []
mean_outs = []
ans_list = []
length_list = []
loss_state = 0
answers = [] # [batch_answers,...]
preds = [] # [batch_preds,...]
preds_trip = []
answers_trip = []
probs = []
ids = []
gumbel_ids = []
uniques_experiment= []
embeddings = np.zeros((len(test_dataset), 300))
idx = 0
all_time = 0
for batch_data in tqdm(test_dataloader):
with torch.no_grad():
qid = batch_data['id']
visual_faetures = torch.tensor(batch_data['img']).float().to(device)
source_seq = tokenizer(batch_data['ques'], padding=True, return_tensors="pt",
add_special_tokens=True).to(device)
input_id = source_seq['input_ids'].to(device)
attention_mask = source_seq['attention_mask'].to(device)
token_type_ids = source_seq['token_type_ids'].to(device)
spatial_feature = torch.tensor(batch_data['spatial']).float().to(device)
most_id = batch_data['mostid']
anchor = model(input_id, attention_mask, token_type_ids, visual_faetures, spatial_feature)
# id = batch_data['id']
# ids.extend(id)
# target = target.squeeze()
# p, idx_1 = torch.topk(target, dim=1, k=1)
# gumbel_ids.extend(idx_1.squeeze().cpu().tolist())
# embeddings[idx: idx + relation.shape[0]] = relation.cpu().numpy()
# idx += relation.shape[0]
anchor = F.normalize(anchor, dim=1, p=2)
if args.embedding:
answer_candidate_tensor_test = model.decode_tail(answer_candidate_tensor)
answer_candidate_tensor_test = F.normalize(answer_candidate_tensor_test, dim=1, p=2)
trip_predict, _ = generate_tripleid(anchor, answer_candidate_tensor_test)
else:
trip_predict, _ = generate_tripleid(anchor, answer_candidate_tensor)
for i, pre in enumerate(most_id):
preds_trip.append(trip_predict[i])
answers_trip.append(most_id[i])
acc_1 = 0
acc_1_trip, ids = cal_acc_multi(answers_trip, preds_trip, return_id=True)
print('epoch %d , acc = %f' % (
epoch, acc_1_trip))
# np.save('best_relation_embedding.npy', embeddings)
# gumbel_dic = {}
# for i in zip(ids, gumbel_ids):
# gumbel_dic[i[0]] = i[1]
# with open('gumbel_visual_train.json','w') as f:
# json.dump(gumbel_dic, f, indent=4)
# for i, id_i in enumerate(ids):
# gumbel_dic[id_i] = embeddings[i]
# with open('train_relation_embedding.pickle','wb') as f:
# pickle.dump(gumbel_dic, f)
if __name__ == "__main__":
test()