/
generate_knowledge.py
163 lines (142 loc) · 5.16 KB
/
generate_knowledge.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
import json
import torch
import pickle
import argparse
from tqdm import tqdm
from pathlib import Path
from transformers import AutoModelForSeq2SeqLM, AutoTokenizer
from utils import calculate_rouge, use_task_specific_params, calculate_bleu_score, trim_batch
def chunks(lst, n):
"""Yield successive n-sized chunks from lst."""
for i in range(0, len(lst), n):
yield lst[i : i + n]
class Comet:
def __init__(self, model_path):
self.device = "cuda" if torch.cuda.is_available() else "cpu"
self.model = AutoModelForSeq2SeqLM.from_pretrained(model_path).to(self.device)
self.tokenizer = AutoTokenizer.from_pretrained(model_path)
task = "summarization"
use_task_specific_params(self.model, task)
self.batch_size = 1
self.decoder_start_token_id = None
def generate(
self,
queries,
decode_method="beam",
num_generate=5,
):
with torch.no_grad():
examples = queries
decs = []
for batch in list(chunks(examples, self.batch_size)):
batch = self.tokenizer(batch, return_tensors="pt", truncation=True, padding="max_length").to(self.device)
input_ids, attention_mask = trim_batch(**batch, pad_token_id=self.tokenizer.pad_token_id)
summaries = self.model.generate(
input_ids=input_ids,
attention_mask=attention_mask,
decoder_start_token_id=self.decoder_start_token_id,
num_beams=num_generate,
num_return_sequences=num_generate,
)
dec = self.tokenizer.batch_decode(summaries, skip_special_tokens=True, clean_up_tokenization_spaces=False)
decs.append(dec)
return decs
all_relations = [
"AtLocation",
"CapableOf",
"Causes",
"CausesDesire",
"CreatedBy",
"DefinedAs",
"DesireOf",
"Desires",
"HasA",
"HasFirstSubevent",
"HasLastSubevent",
"HasPainCharacter",
"HasPainIntensity",
"HasPrerequisite",
"HasProperty",
"HasSubEvent",
"HasSubevent",
"HinderedBy",
"InheritsFrom",
"InstanceOf",
"IsA",
"LocatedNear",
"LocationOfAction",
"MadeOf",
"MadeUpOf",
"MotivatedByGoal",
"NotCapableOf",
"NotDesires",
"NotHasA",
"NotHasProperty",
"NotIsA",
"NotMadeOf",
"ObjectUse",
"PartOf",
"ReceivesAction",
"RelatedTo",
"SymbolOf",
"UsedFor",
"isAfter",
"isBefore",
"isFilledBy",
"oEffect",
"oReact",
"oWant",
"xAttr",
"xEffect",
"xIntent",
"xNeed",
"xReact",
"xReason",
"xWant",
]
def get_knowledge(model, dataset):
knowledge_set = []
relation_set = ["oEffect", "oReact", "oWant", "xAttr", "xEffect",
"xIntent", "xNeed", "xReact", "xReason", "xWant"]
# relation_set = ["oReact", "xReact"]
count = 1
for conv in dataset:
conv_knowledge = []
for utter in conv:
print(f'{count} processed')
utter_knowledge = {}
# utter_knowledge = 'this person feels'
queries = []
utterance = utter['utterance']
for r in relation_set:
query = "{} {} [GEN]".format(utterance, r)
queries.append(query)
results = model.generate(queries, decode_method="beam", num_generate=5)
for relation, result in zip(relation_set, results):
utter_knowledge[relation] = ' ==sep== '.join(result)
# if relation == 'oReact':
# utter_knowledge = utter_knowledge + result[0] + ' and' + result[1] + ' . '
# else:
# utter_knowledge = utter_knowledge + 'others feels' + result[0] + ' and ' + result[1] + ' . '
conv_knowledge.append(utter_knowledge)
count += 1
knowledge_set.append(conv_knowledge)
return knowledge_set
if __name__ == "__main__":
# sample usage
print("model loading ...")
comet = Comet("./comet-atomic_2020_BART")
comet.model.zero_grad()
print("model loaded")
train_data = pickle.load(open('/data2/ljn/RECCON-ERC/dd_data/dailydialog_train.pkl', 'rb'), encoding='latin1')
dev_data = pickle.load(open('/data2/ljn/RECCON-ERC/dd_data/dailydialog_dev.pkl', 'rb'), encoding='latin1')
test_data = pickle.load(open('/data2/ljn/RECCON-ERC/dd_data/dailydialog_test.pkl', 'rb'), encoding='latin1')
print('train data')
train_knowledge = get_knowledge(comet, train_data)
print('dev data')
dev_knowledge = get_knowledge(comet, dev_data)
print('test data')
test_knowledge = get_knowledge(comet, test_data)
pickle.dump(train_knowledge, open('/data2/ljn/RECCON-ERC/dd_data/dailydialog_train_know.pkl', 'wb'))
pickle.dump(dev_knowledge, open('/data2/ljn/RECCON-ERC/dd_data/dailydialog_dev_know.pkl', 'wb'))
pickle.dump(test_knowledge, open('/data2/ljn/RECCON-ERC/dd_data/dailydialog_test_know.pkl', 'wb'))