-
Notifications
You must be signed in to change notification settings - Fork 0
/
valid_modeltts.py
140 lines (115 loc) · 4.53 KB
/
valid_modeltts.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
import argparse
import seaborn
import torch
import matplotlib.pyplot as plt
import random
from dataset.vocab import WordVocab
import pdb
from dataset_multi import Dataset
from torch.utils.data import DataLoader
import yaml
def random_word(sentence, vocab):
tokens = sentence.split()
tokens_len = [len(token) for token in tokens]
chars = [char for char in sentence]
output_label = []
for i, char in enumerate(chars):
prob = random.random()
if prob < 0.15:
prob /= 0.15
# 80% randomly change token to mask token
if prob < 0.8:
chars[i] = vocab.mask_index
# 10% randomly change token to random token
elif prob < 0.9:
chars[i] = random.randrange(vocab.vocab_size)
# 10% randomly change token to current token
else:
chars[i] = vocab.char2index(char)
output_label.append(vocab.char2index(char))
else:
chars[i] = vocab.char2index(char)
output_label.append(0)
return chars, output_label
def draw(data, x, y, ax):
seaborn.heatmap(data,
xticklabels=x, square=True, yticklabels=y, vmin=0.0, vmax=1.0, # 取值0-1
cbar=False, ax=ax)
def Modelload(path):
assert path is not None
print(f"path:{path}")
mlm_encoder = torch.load(path)
return mlm_encoder
# 验证模型是否收敛
def main():
# parser = argparse.ArgumentParser()
#
# parser.add_argument("-m", "--model_path", required=True, type=str, help="model of pretrain")
# parser.add_argument("-v", "--vocab_path", required=True, type=str, help="path of vocab")
# args = parser.parse_args()
#
# model_path = args.model_path
# vocab_path = args.vocab_path
#
# vocab = WordVocab.load_vocab(vocab_path)
#
# model = torch.load(model_path,'cpu')
# model = model.to('cuda')
# model.eval()
#
# sent = '_I _l _o _v _e _C _h _i _n _a _!'.split()
#
# text = 'I love China!'
# sent1, label = random_word(text, vocab)
# position = [*range(13)]
# sent1 = torch.tensor(sent1).long().unsqueeze(0).to('cuda')
# position1 = torch.tensor(position).long().unsqueeze(0).to('cuda')
# mask_lm_output, attn_list = model.module.forward(sent1, position1)
# mask_lm_output = torch.argmax(mask_lm_output,dim=2)
# pdb.set_trace()
# print(mask_lm_output)
parser = argparse.ArgumentParser()
parser.add_argument("-m", "--model_path", required=True, type=str, help="model of pretrain")
parser.add_argument("-v", "--vocab_path", required=True, type=str, help="path of vocab")
args = parser.parse_args()
model_path = args.model_path
vocab_path = args.vocab_path
model = torch.load(model_path,'cpu')
model = model.to('cuda')
model.eval()
path = "LibriTTS_StyleSpeech_multilingual_diffusion_style_3layer"
# path = "VNTTS"
# path = "LibriTTS_StyleSpeech_multilingual_diffusion_style_EN"
preprocess_config = yaml.load(
open("./config/config_kaga/{0}/preprocess.yaml".format(path), "r"), Loader=yaml.FullLoader
)
train_config = yaml.load(
open("./config/config_kaga/{0}/train.yaml".format(path), "r"), Loader=yaml.FullLoader
)
val_dataset = Dataset("val.txt", preprocess_config, train_config, sort=False, drop_last=False, random=False)
valid_data_loader = DataLoader(val_dataset, batch_size=4, shuffle=False, num_workers=0, collate_fn=val_dataset.collate_fn)
predict = 0
total = 0
for batch in valid_data_loader:
input_ids = batch["mlm_input"]
position = batch["input_position"]
labels = batch["mlm_label"]
src_masks = batch["src_masks"]
with torch.no_grad():
outputs, attn_list = model(input_ids, src_masks)
outputs = torch.argmax(outputs, dim=2)
for output, label, input_id in zip(outputs.detach().cpu(), labels.detach().cpu(), input_ids.detach().cpu()):
mask_index = (input_id == 4).nonzero(as_tuple=True)[0]
output_index = torch.index_select(output, 0, mask_index)
label_index = torch.index_select(label, 0, mask_index)
print("output: ", output_index)
print("label: ", label_index)
print("*" * 20)
predict += torch.eq(output_index, label_index).sum()
total += label_index.shape[0]
print(f"True Predict / Total: {predict} / {total}")
acc = predict / total
acc = round(acc.item()*100, 3)
print(f"Acc: {acc} % ")
if __name__ == '__main__':
main()