-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathevaluate.py
198 lines (173 loc) · 8.39 KB
/
evaluate.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
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
import torch
import random
from data import loadPreparedData,Dictionary
from config import SOS_token,EOS_token,MAXLEN
from utils import sentence2index,load_embeddings
from model import *
from collections import OrderedDict
MAX_LENGTH = MAXLEN
indexesFromSentence = sentence2index
USE_CUDA = torch.cuda.is_available()
device = torch.device("cuda" if USE_CUDA else "cpu")
"""
some refer to：https://github.com/ywk991112/pytorch-chatbot/blob/master/evaluate.py
"""
def normalizeString(s):
# lower
s = unicodeToAscii(s.lower().strip())
# special rules to deal with . ! ?
s = re.sub(r"([.!?])", r" \1", s)
# filter special symbol with whitespaces
s = re.sub(r"[^a-zA-Z.!?]+", r" ", s)
# multiple whitespaces to single whitespace
s = re.sub(r"\s+", r" ", s).strip()
return s
class Sentence:
def __init__(self, decoder_hidden, last_idx=SOS_token, sentence_idxes=[], sentence_scores=[]):
if(len(sentence_idxes) != len(sentence_scores)):
raise ValueError("length of indexes and scores should be the same")
self.decoder_hidden = decoder_hidden
self.last_idx = last_idx
self.sentence_idxes = sentence_idxes
self.sentence_scores = sentence_scores
def avgScore(self):
if len(self.sentence_scores) == 0:
raise ValueError("Calculate average score of sentence, but got no word")
# return mean of sentence_score
return sum(self.sentence_scores) / len(self.sentence_scores)
def addTopk(self, topi, topv, decoder_hidden, beam_size, voc):
topv = torch.log(topv)
terminates, sentences = [], []
for i in range(beam_size):
if topi[0][i] == EOS_token:
terminates.append(([voc.index2word[idx.item()] for idx in self.sentence_idxes] + ['<EOS>'],
self.avgScore())) # tuple(word_list, score_float
continue
idxes = self.sentence_idxes[:] # pass by value
scores = self.sentence_scores[:] # pass by value
idxes.append(topi[0][i])
scores.append(topv[0][i])
sentences.append(Sentence(decoder_hidden, topi[0][i], idxes, scores))
return terminates, sentences
def toWordScore(self, voc):
words = []
for i in range(len(self.sentence_idxes)):
if self.sentence_idxes[i] == EOS_token:
words.append('<EOS>')
else:
words.append(voc.index2word[self.sentence_idxes[i].item()])
if self.sentence_idxes[-1] != EOS_token:
words.append('<EOS>')
return (words, self.avgScore())
def beam_decode(decoder, decoder_hidden, encoder_outputs, keyword, voc, beam_size, max_length=MAX_LENGTH):
terminal_sentences, prev_top_sentences, next_top_sentences = [], [], []
prev_top_sentences.append(Sentence(decoder_hidden))
for i in range(max_length):
for sentence in prev_top_sentences:
decoder_input = torch.LongTensor([[sentence.last_idx]])
decoder_input = decoder_input.to(device)
decoder_hidden = sentence.decoder_hidden
decoder_output, decoder_hidden, _ = decoder(encoder_outputs,decoder_input, keyword, decoder_hidden)
topv, topi = decoder_output.topk(beam_size)
term, top = sentence.addTopk(topi, topv, decoder_hidden, beam_size, voc)
terminal_sentences.extend(term)
next_top_sentences.extend(top)
next_top_sentences.sort(key=lambda s: s.avgScore(), reverse=True)
prev_top_sentences = next_top_sentences[:beam_size]
next_top_sentences = []
terminal_sentences += [sentence.toWordScore(voc) for sentence in prev_top_sentences]
terminal_sentences.sort(key=lambda x: x[1], reverse=True)
n = min(len(terminal_sentences), 15)
return terminal_sentences[:n]
def decode(decoder,decoder_hidden,encoder_outputs,keyword,voc,max_length=MAX_LENGTH):
decoder_input = torch.LongTensor([[SOS_token]])
decoder_input = decoder_input.to(device)
decoded_words = []
decoder_attentions = torch.zeros(max_length, max_length) #TODO: or (MAX_LEN+1, MAX_LEN+1)
for di in range(max_length):
decoder_output, decoder_hidden = decoder(
encoder_outputs,decoder_input,keyword,decoder_hidden
)
_, topi = decoder_output.topk(3)
ni = topi[0][0]
if ni == EOS_token:
decoded_words.append('<EOS>')
break
else:
decoded_words.append(voc.index2word[ni.item()])
decoder_input = torch.LongTensor([[ni]])
decoder_input = decoder_input.to(device)
return decoded_words, decoder_attentions[:di + 1]
def evaluate(encoder,decoder,src_voc,tag_voc,sentence,keyword,beam_size,max_length=MAX_LENGTH):
indexes_batch = [indexesFromSentence(src_voc, sentence)]
lengths = [len(indexes) for indexes in indexes_batch]
keyword_batch = [tag_voc.word2index[keyword]]
input_batch = torch.LongTensor(indexes_batch)
input_batch = input_batch.to(device)
keyword_batch = torch.LongTensor(keyword_batch)
keyword_batch = keyword_batch.to(device)
encoder_outputs, (encoder_h,encoder_c) = encoder(input_batch,lengths,None)
decoder_hidden = (encoder_h[:decoder.n_layers],encoder_c[:decoder.n_layers])
if beam_size == 1:
return decode(decoder, decoder_hidden, encoder_outputs, keyword_batch, tag_voc)
else:
return beam_decode(decoder, decoder_hidden, encoder_outputs, keyword_batch, tag_voc, beam_size)
# def evaluateRandomly(encoder, decoder, voc, pairs, reverse, beam_size, n=10):
# for _ in range(n):
# pair = random.choice(pairs)
# print("=============================================================")
# if reverse:
# print('>', " ".join(reversed(pair[0].split())))
# else:
# print('>', pair[0])
# if beam_size == 1:
# output_words, _ = evaluate(encoder, decoder, voc, pair[0], beam_size)
# output_sentence = ' '.join(output_words)
# print('<', output_sentence)
# else:
# output_words_list = evaluate(encoder, decoder, voc, pair[0], beam_size)
# for output_words, score in output_words_list:
# output_sentence = ' '.join(output_words)
# print("{:.3f} < {}".format(score, output_sentence))
def evaluateInput(encoder,decoder,src_voc,tag_voc,beam_size):
question = ''
while(1):
try:
question = input('q: ')
keyword = input('k: ')
if question == 'q': break
if beam_size == 1:
output_words, _ = evaluate(encoder,decoder,src_voc,tag_voc,question,keyword,beam_size)
output_sentence = ' '.join(output_words)
print('a:', output_sentence)
else:
output_words_list = evaluate(encoder,decoder,src_voc,tag_voc,question,keyword,beam_size)
for output_words, score in output_words_list:
output_sentence = ' '.join(output_words)
print("{:.3f} < {}".format(score, output_sentence))
except KeyError:
print("Incorrect spelling.")
def runTest(n_layers,input_size,hidden_size,modelFile,beam_size,corpus_name):
checkpoint = None
# load word parameter
checkpoint = torch.load(modelFile,map_location='cpu')
src_voc = checkpoint['src_dict']
tag_voc = checkpoint['tag_dict']
source_embedding = nn.Embedding(src_voc.n_words,input_size)
target_embedding = nn.Embedding(tag_voc.n_words,input_size)
source_embedding.load_state_dict(checkpoint['src_embedding'])
target_embedding.load_state_dict(checkpoint['tag_embedding'])
#build
encoder = EncoderRNN(source_embedding,input_size,hidden_size,n_layers,bidirectional=True,dropout=0.1,rnn_type='LSTM')
decoder = DecoderRNN(target_embedding,input_size,hidden_size,tag_voc.n_words,n_layers,rnn_type='LSTM',use_ATTN=True,dropout=0.1)
# load model
encoder.load_state_dict(checkpoint['encoder'])
decoder.load_state_dict(checkpoint['decoder'])
encoder.train(False)
decoder.train(False)
# to GPU
encoder = encoder.to(device)
decoder = decoder.to(device)
torch.set_grad_enabled(False)
#evaluate
evaluateInput(encoder,decoder,src_voc,tag_voc,beam_size)