-
Notifications
You must be signed in to change notification settings - Fork 2
/
Copy pathsubmit_result.py
286 lines (274 loc) · 11 KB
/
submit_result.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
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
# import os
# import sys
# import glob
# import json
# # sys.path.append("code/TorchBlocks/")
# # from torchblocks.utils.options import Argparser
# # from torchblocks.core.utils import is_apex_available
# sys.path.append("code/")
# from packages import Argparser, is_apex_available
# # WARNING: 统一采用绝对路径！！
# def pred_BIO(path_word: str, path_sample: str, batch_size: int = 1,
# # model_path="/home/mw/project/data/best_model/gmodel_spancls_20220505065060", # 0.8105264749920598
# # model_path="/home/mw/project/data/best_model/gmodel_spancls_202205077213", # 0.8132664117368327
# # model_path="/home/mw/project/data/best_model/gmodel_gp_20220508xxxx", # 0.8135973553679232
# # model_path="/home/mw/project/data/best_model/gmodel_gpswa_202205089550", # 0.8135456792962686
# # model_path="/home/mw/project/data/best_model/gmodel_spancls_20220509016006", # 0.8131485592934022
# # model_path="/home/mw/project/data/best_model/gmodel_spancls_202205101483", # 0.8144165048602052
# # model_path="/home/mw/project/data/best_model/gmodel_spancls_2022051219618", # 0.8154596708823378
# # model_path="/home/mw/project/data/best_model/gmodel_gp_2022051235705", # 0.814864726037745
# # model_path="/home/mw/project/data/best_model/gmodel_spancls_2022051428128", # 0.8153485305142973
# # model_path="/home/mw/project/data/best_model/gmodel_spancls_2022051515130", # 0.8153156273245536
# # model_path="/home/mw/project/data/best_model/gmodel_spancls_2022051528877", # 0.8154189358838884
# # model_path="/home/mw/project/data/best_model/gmodel_spancls_2022051617829", # 0.8104775463912551
# model_path="/home/mw/project/data/best_model/gmodel_spancls_2022051629391", # 0.815385819848358
# submit_result_file="/home/mw/project/results.txt"
# ):
# basename, ext = os.path.splitext(os.path.basename(path_word))
# model_path = os.path.abspath(model_path)
# print(model_path); os.system("ls %s" % model_path)
# json_file = glob.glob(os.path.join(model_path, "*_opts.json"))[0]
# opts = Argparser.parse_args_from_json(json_file=json_file)
# opts.output_dir = model_path
# opts.pretrained_model_path = model_path
# opts.checkpoint_predict_code = "" # 线上只有一级目录
# opts.do_train = False
# opts.do_eval = False
# opts.do_predict = True
# opts.per_gpu_test_batch_size = batch_size
# opts.gradient_accumulation_steps = 1
# opts.data_dir = "../data/tmp_data/predict/"
# opts.test_input_file = f"{basename}.jsonl"
# opts.fp16= False
# json_file = os.path.join(model_path, "predict_opts.json")
# with open(str(json_file), 'w') as f:
# json.dump(vars(opts), f, ensure_ascii=False, indent=4)
# # if not is_apex_available():
# # cmd = \
# # """
# # # git clone https://github.com/NVIDIA/apex
# # cd apex
# # pip install -v --disable-pip-version-check --no-cache-dir ./
# # """
# # print(cmd)
# # os.system(cmd)
# path_word = os.path.abspath(path_word)
# cmd = \
# """
# # sh init.sh
# cd code/
# python prepare_data.py \
# --version=predict \
# --labeled_files \
# ../data/contest_data/train_data/train.txt \
# --test_files %s \
# --output_dir=../data/tmp_data/ \
# --n_splits=1 \
# --seed=42
# python run_span_classification_v1.py \
# %s
# """ % (path_word, json_file)
# print(cmd)
# os.system(cmd)
# checkpoint_path = os.path.join(model_path, opts.checkpoint_predict_code)
# result_file_path = glob.glob(os.path.join(checkpoint_path, "*.predictions.txt"))[0]
# cmd = \
# """
# cp %s %s
# """ % (result_file_path, submit_result_file)
# print(cmd)
# os.system(cmd)
# # if __name__ == "__main__":
# # pred_BIO(
# # "data/contest_data/preliminary_test_b/word_per_line_preliminary_B.txt",
# # "data/contest_data/preliminary_test_b/sample_per_line_preliminary_B.txt",
# # submit_result_file="results.txt",
# # )
# ---------------------------------------------------------------------------------------------
import os
import glob
import json
# WARNING: 统一采用绝对路径！！
def pred_BIO(path_word: str, path_sample: str, batch_size: int = 1,
# model_path="/home/mw/project/best_model",
# model_path="/home/mw/project/data/best_model/gmodel_gpv2_2022051811505", # 0.8176488551464742
# model_path="/home/mw/project/data/best_model/gmodel_gpv2_2022051916110", # 0.8176612608365892
# model_path="/home/mw/project/data/best_model/gmodel_gpv2_2022052133414", # 0.8170789177954735
model_path="/home/mw/project/data/best_model/gmodel_gpv2_2022052125903", # 0.8173645640130256
submit_result_file="/home/mw/project/results.txt"
):
model_path = os.path.abspath(model_path)
print(model_path); os.system("ls %s" % model_path)
# cmd = \
# f"""
# python exp_gaiic_global_pointer_v2.py \
# --experiment_code=experiment_bert_base_fold0_gp_v2_pre_v62 \
# --task_name=gaiic \
# --model_type=nezha \
# --do_lower_case \
# --pretrained_model_path={model_path} \
# --data_dir=./ \
# --output_dir=./ \
# --do_predict_test \
# --test_input_file={path_sample} \
# --eval_checkpoint_path={model_path} \
# --submit_file_path={submit_result_file} \
# --evaluate_during_training \
# --train_max_seq_length=128 \
# --eval_max_seq_length=128 \
# --test_max_seq_length=128 \
# --per_gpu_train_batch_size=16 \
# --per_gpu_eval_batch_size=32 \
# --per_gpu_test_batch_size={batch_size} \
# --gradient_accumulation_steps=1 \
# --learning_rate=3e-5 \
# --other_learning_rate=1e-3 \
# --weight_decay=0.001 \
# --scheduler_type=cosine \
# --base_model_name=bert \
# --warmup_proportion=0.1 \
# --max_grad_norm=1.0 \
# --num_train_epochs=10 \
# --use_rope \
# --do_lstm \
# --num_lstm_layers=2 \
# --adam_epsilon=1e-8 \
# --post_lstm_dropout=0.5 \
# --inner_dim=64 \
# --loss_type=pcl \
# --do_awp \
# --do_rdrop \
# --seed=42
# """
cmd = \
f"""
cd code/
python exp_gaiic_global_pointer_v2.py \
--experiment_code=experiment_bert_base_fold0_gp_v2_pre_v62 \
--task_name=gaiic \
--model_type=nezha \
--do_lower_case \
--pretrained_model_path={model_path} \
--data_dir=/home/mw/temp/10_folds_data/ \
--train_input_file=train.all.jsonl \
--eval_input_file=dev.0.jsonl \
--output_dir=../data/model_data/ \
--do_predict_test \
--save_best \
--test_input_file={path_sample} \
--eval_checkpoint_path={model_path} \
--submit_file_path={submit_result_file} \
--evaluate_during_training \
--train_max_seq_length=128 \
--eval_max_seq_length=128 \
--test_max_seq_length=128 \
--per_gpu_train_batch_size=16 \
--per_gpu_eval_batch_size=32 \
--per_gpu_test_batch_size={batch_size} \
--gradient_accumulation_steps=1 \
--learning_rate=3e-5 \
--other_learning_rate=1e-3 \
--weight_decay=0.001 \
--scheduler_type=cosine \
--base_model_name=bert \
--warmup_proportion=0.1 \
--max_grad_norm=1.0 \
--num_train_epochs=10 \
--use_rope \
--do_lstm \
--do_fgm \
--num_lstm_layers=2 \
--adam_epsilon=1e-8 \
--post_lstm_dropout=0.5 \
--inner_dim=64 \
--loss_type=pcl \
--pcl_epsilon=2.5 \
--pcl_alpha=1.5 \
--do_awp \
--awp_f1=0.810 \
--awp_lr=0.1 \
--do_rdrop \
--rdrop_weight=0.4 \
--rdrop_epoch=1 \
--seed=42
"""
print(cmd)
os.system(cmd)
if __name__ == "__main__":
pred_BIO(
path_word= "",
path_sample="./test_submit_dev_0.txt",
model_path='./best_model',
submit_result_file="tmp_results.txt",
)
dev_0 = './dev.0.jsonl'
import json
from utils import get_entity_biob
true = []
with open(dev_0) as fr:
for line in fr.readlines():
line = json.loads(line)
true.append(line)
sentences = []
sentence_counter = 0
with open('tmp_results.txt', encoding="utf-8") as f:
lines = f.readlines()
current_words = []
current_labels = []
for row in lines:
row = row.rstrip("\n")
if row != "":
token, label = row[0], row[2:]
current_words.append(token)
current_labels.append(label)
else:
if not current_words:
continue
assert len(current_words) == len(current_labels), "word len doesn't match label length"
sentence = {
"id": str(sentence_counter),
"tokens": current_words,
"ner_tags": current_labels
}
sentence_counter += 1
current_words = []
current_labels = []
sentences.append(sentence)
result = []
for x,y in zip(true,sentences):
true_ents = get_entity_biob(x['ner_tags'],None)
pred_ents = get_entity_biob(y['ner_tags'],None)
x['true'] = true_ents
x['pred'] = pred_ents
assert "".join(x['tokens']) == "".join(y['tokens'])
result.append(x)
import torch
import numpy as np
class MetricsCalculator(object):
def __init__(self):
super().__init__()
def get_sample_f1(self, y_pred, y_true):
y_pred = torch.gt(y_pred, 0).float()
return 2 * torch.sum(y_true * y_pred) / torch.sum(y_true + y_pred)
def get_sample_precision(self, y_pred, y_true):
y_pred = torch.gt(y_pred, 0).float()
return torch.sum(y_pred[y_true == 1]) / (y_pred.sum() + 1)
def get_evaluate_fpr(self, y_pred):
pred = []
true = []
for i, x in enumerate(y_pred):
p = x['pred']
for pp in p:
pred.append((i, pp[0], pp[1], pp[2]))
t = x['true']
for tt in t:
true.append((i, tt[0], tt[1], tt[2]))
R = set(pred)
T = set(true)
X = len(R & T)
Y = len(R)
Z = len(T)
f1, precision, recall = 2 * X / (Y + Z), X / Y, X / Z
return f1, precision, recall
m = MetricsCalculator()
m.get_evaluate_fpr(result)