-
Notifications
You must be signed in to change notification settings - Fork 0
/
ranker.py
147 lines (131 loc) · 6.48 KB
/
ranker.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
import warnings
warnings.simplefilter(action='ignore', category=FutureWarning)
import torch
from transformers import T5Tokenizer, T5Config, T5ForConditionalGeneration, GPT2Tokenizer, GPT2LMHeadModel
from scipy.special import softmax
import numpy
torch.manual_seed(0)
DEVICE = torch.device('cuda:0' if torch.cuda.is_available() else 'cpu')
"""
THIS FILE MAINLY HANDLES THE RANKER RELATED WORK
"""
class T5:
def __init__(self, model_dir):
self.name = "T5"
self.tokenizer = T5Tokenizer.from_pretrained('t5-base')
config = T5Config.from_pretrained('t5-base')
self.model = T5ForConditionalGeneration.from_pretrained(
model_dir, from_tf=True, config=config)
self.model.eval()
self.model.to(DEVICE)
self.softmax = torch.nn.Softmax(dim=-1).to(DEVICE)
def predict(self, document, query, conf):
if "CHUNK_SIZE" in conf:
scores = []
segments = document.split(" ")
x = 0
y = len(segments)
for i in range(x, y, conf["CHUNK_SIZE"]):
x = i
temp = ' '.join(segments[x:x + conf["CHUNK_SIZE"]])
temp += ' </s>'
prob = []
encoder_input_ids = self.tokenizer.encode(document, return_tensors='pt').to(DEVICE)
decoder_input_ids = self.tokenizer.encode(query, return_tensors='pt').to(DEVICE)
with torch.no_grad():
outputs = self.model(input_ids=encoder_input_ids, labels=decoder_input_ids).to(DEVICE)
logits = outputs[1][0]
distributions = softmax(logits.numpy(), axis=1)
for index, val in enumerate(decoder_input_ids[0]):
prob.append(distributions[index][val])
score = numpy.sum(numpy.log10(prob))
scores.append(score)
return max(scores)
else:
document += ' </s>'
prob = []
encoder_input_ids = self.tokenizer.encode(document, return_tensors='pt').to(DEVICE)
decoder_input_ids = self.tokenizer.encode(query, return_tensors='pt').to(DEVICE)
with torch.no_grad():
outputs = self.model(input_ids=encoder_input_ids, labels=decoder_input_ids)
logits = outputs[1][0]
distributions = softmax(logits.numpy(), axis=1)
for index, val in enumerate(decoder_input_ids[0]):
prob.append(distributions[index][val])
score = numpy.sum(numpy.log10(prob))
return score
def tokenize(self, text):
encoded_inputs = self.tokenizer(text, padding=True, truncation=True, return_tensors="pt").to(DEVICE)
return encoded_inputs
def batchPredict(self, documents, decoder_inputs, conf):
documents = [document + ' </s>' for document in documents]
# querys = [query] * len(documents)
encoder_inputs = self.tokenizer(documents, padding=True, truncation=True, return_tensors="pt").to(
DEVICE)
# encoded_decoder_inputs = self.tokenizer(querys, padding=True, truncation=True, return_tensors="pt").to(DEVICE)
decoder_input_ids = decoder_inputs["input_ids"]
# scores = []
with torch.no_grad():
outputs = self.model(input_ids=encoder_inputs["input_ids"],
labels=decoder_input_ids,
attention_mask=encoder_inputs["attention_mask"])
batch_logits = outputs[1] # shape(batch_size, decoder_dim, num_tokens)
# batch_logits = batch_logits.cpu()
distributions = self.softmax(batch_logits) # shape(batch_size, decoder_dim, num_tokens)
decoder_input_ids = decoder_input_ids.unsqueeze(-1) # shape(batch_size, decoder_dim, 1)
batch_probs = torch.gather(distributions, 2, decoder_input_ids).squeeze(-1) # shape(batch_size, decoder_dim)
masked_log_probs = torch.log10(batch_probs) # shape(batch_size, decoder_dim)
scores = torch.sum(masked_log_probs, 1) # shape(batch_size)
# for logits in batch_logits:
# # distributions = softmax(logits.numpy(), axis=1)
# distributions = self.softmax(logits)
# prob = []
# for index, val in enumerate(decoder_input_ids):
# prob.append(distributions[index][val])
# score = numpy.sum(numpy.log10(prob))
# scores.append(score)
return scores
class GPT2:
def __init__(self, model_dir=None):
self.name = "GPT2"
if model_dir is not None:
pass
else:
self.tokenizer = GPT2Tokenizer.from_pretrained('gpt2-large')
self.model = GPT2LMHeadModel.from_pretrained('gpt2-large')
self.model.to(DEVICE)
def predict(self, document, query, conf):
if "CHUNK_SIZE" in conf:
scores = []
segments = document.split(" ")
x = 0
y = len(segments)
for i in range(x, y, conf["CHUNK_SIZE"]):
x = i
temp = ' '.join(segments[x:x + conf["CHUNK_SIZE"]])
prob = []
input_ids = self.tokenizer.encode(temp + " " + query, return_tensors='pt').to(DEVICE)
query_ids = self.tokenizer.encode(query).to(DEVICE)
doc_length = len(input_ids[0]) - len(query_ids)
with torch.no_grad():
outputs = self.model(input_ids=input_ids)
logits = outputs[0][0]
distributions = softmax(logits.numpy(), axis=1)
for index, val in enumerate(query_ids):
prob.append(distributions[doc_length + index][val])
score = numpy.sum(numpy.log10(prob))
scores.append(score)
return max(scores)
else:
prob = []
input_ids = self.tokenizer.encode(document + " " + query, return_tensors='pt').to(DEVICE)
query_ids = self.tokenizer.encode(query).to(DEVICE)
doc_length = len(input_ids[0]) - len(query_ids)
with torch.no_grad():
outputs = self.model(input_ids=input_ids)
logits = outputs[0][0]
distributions = softmax(logits.numpy(), axis=1)
for index, val in enumerate(query_ids):
prob.append(distributions[doc_length + index][val])
score = numpy.sum(numpy.log10(prob))
return score