-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathtrain.py
195 lines (171 loc) · 9.06 KB
/
train.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
import sys
sys.path.extend(["../utils", ".."])
from model import EntityTypeClassification
from transformers import AdamW, get_linear_schedule_with_warmup
from evaluation import model_eval
from data_util import Entity_Dataset, my_collate_fn
from tqdm import tqdm, trange
import torch
from torch import nn
import time
import os
import numpy as np
import argparse
from tensorboardX import SummaryWriter
from torch.utils.data import DataLoader
from load_cleanterms import CLEANTERMS
def run(args):
writer = SummaryWriter(comment='TypeClassificationWithMask')
cleanterms = CLEANTERMS(args.clean_term_path)
train_control_dict = {'max_sgr_n':1, 'no_short_upper':True, 'only_one':True, 'min_entity_word_count':1, 'max_sty_n':3}
single_eval_control_dict = {'max_sty_n':1, 'no_short_upper':True}
multi_eval_control_dict = {'min_sty_n':2, 'no_short_upper':True}
train_dataset = Entity_Dataset(args.train_text_file, args.train_match_file, cleanterms, tokenizer_name=args.model_name_or_path, mask_ratio=args.mask_ratio, control_dict=train_control_dict, debug=args.debug)
single_eval_dataset = Entity_Dataset(args.eval_text_file, args.eval_match_file, cleanterms, args.model_name_or_path, mask_ratio=0.0, control_dict=single_eval_control_dict)
multi_eval_dataset = Entity_Dataset(args.eval_text_file, args.eval_match_file, cleanterms, args.model_name_or_path, mask_ratio=0.0, control_dict=multi_eval_control_dict)
train_dataloader = DataLoader(train_dataset, batch_size=args.train_batch_size, shuffle=True, collate_fn=my_collate_fn, num_workers=args.num_workers)
test_dataloader = DataLoader(single_eval_dataset, batch_size=args.train_batch_size, shuffle=False, collate_fn=my_collate_fn, num_workers=args.num_workers)
test_multi_dataloader = DataLoader(multi_eval_dataset, batch_size=args.train_batch_size, shuffle=False, collate_fn=my_collate_fn, num_workers=args.num_workers)
#if args.output_dir == "":
# args.output_dir = f"output/lr{args.learning_rate}_mask{args.mask_ratio}"
# if args.debug:
# args.output_dir = args.output_dir + "_debugAlpha"
args.output_dir = args.output_dir + f"_lr{args.learning_rate}_minwordcount-1_maxstyn-3"
try:
os.system(f"rm -rf {args.output_dir}")
except BaseException:
pass
if not os.path.exists(args.output_dir):
os.makedirs(args.output_dir)
if not os.path.exists(os.path.join(args.output_dir, "model")):
os.makedirs(os.path.join(args.output_dir, "model"))
if not os.path.exists(os.path.join(args.output_dir, "eval")):
os.makedirs(os.path.join(args.output_dir, "eval"))
# model & optimizer & scheduler
train_steps = int(args.train_epoch * len(train_dataloader))
model = EntityTypeClassification(args.model_name_or_path, len(train_dataset.umls_labels)).to(args.device)
no_decay = ["bias", "LayerNorm.weight"]
optimizer_grouped_parameters = [
{
"params": [p for n, p in model.named_parameters() if not any(nd in n for nd in no_decay)],
"weight_decay": args.weight_decay,
},
{"params": [p for n, p in model.named_parameters() if any(
nd in n for nd in no_decay)], "weight_decay": 0.0},
]
optimizer = AdamW(optimizer_grouped_parameters,
lr=args.learning_rate, eps=args.adam_epsilon)
scheduler = get_linear_schedule_with_warmup(
optimizer, num_warmup_steps=int(0.1 * train_steps), num_training_steps=train_steps
)
global_step = 0
for train_epoch_idx in range(args.train_epoch):
model.train()
epoch_iterator = tqdm(train_dataloader, desc="Iteration", ascii=True)
epoch_loss = 0.
for batch_idx, batch in enumerate(epoch_iterator):
sentence = torch.LongTensor(batch[0]).to(args.device)
entity_mark = torch.LongTensor(batch[1]).to(args.device)
labels = torch.LongTensor(batch[2]).to(args.device)
loss, _ = model(sentence, entity_mark, labels.argmax(-1))
batch_loss = loss.item()
epoch_loss += batch_loss
writer.add_scalar('batch_loss', loss, global_step=global_step)
if args.gradient_accumulation_steps > 1:
loss = loss / args.gradient_accumulation_steps
loss.backward()
epoch_iterator.set_description("Epoch_loss: %s, Batch_loss: %0.4f" %
(epoch_loss / (batch_idx + 1), batch_loss))
if (global_step + 1) % args.gradient_accumulation_steps == 0:
torch.nn.utils.clip_grad_norm_(
model.parameters(), args.max_grad_norm)
optimizer.step()
scheduler.step() # Update learning rate schedule
model.zero_grad()
global_step += 1
if global_step % args.save_step == 0 and global_step > 0:
save_path = os.path.join(
args.output_dir, 'model', f'model_{global_step}.pth')
torch.save(model, save_path)
if global_step % args.eval_step == 0 and global_step > 0:
test_eval_dict = model_eval(args, model, test_dataloader, eval=True, predict=True, tokenizer=train_dataset.tokenizer,
eval_path=os.path.join(args.output_dir, 'single_eval_log.txt'),
predict_path=os.path.join(args.output_dir, 'eval', f'test-{global_step}.csv'),
output_suffix=f"Global step:{global_step}")
multi_eval_dict = model_eval(args, model, test_multi_dataloader, eval=True, predict=True, tokenizer=train_dataset.tokenizer,
eval_path=os.path.join(args.output_dir, 'multi_eval_log.txt'),
predict_path=os.path.join(args.output_dir, 'eval', f'multi-test-{global_step}.csv'))
for key, value in test_eval_dict.items():
writer.add_scalar(key, value, global_step=global_step)
for key, value in multi_eval_dict.items():
writer.add_scalar("mulit_" + key, value, global_step=global_step)
print("single_eval_dict: ", test_eval_dict)
print("multi_eval_dict: ", multi_eval_dict)
save_path = os.path.join(args.output_dir, 'model', 'last.pth')
torch.save(model, save_path)
print(f'Save model to {save_path} successful.')
def main():
parser = argparse.ArgumentParser()
parser.add_argument('--train_text_file', type=str, default='example/data/corpus.txt')
parser.add_argument('--train_match_file', type=str, default='example/data/corpus_tagged.txt')
parser.add_argument('--eval_text_file', type=str, default='/raid/zheng/type_data/dev_text.txt')
parser.add_argument('--eval_match_file', type=str, default='/raid/zheng/type_data/dev_match.txt')
parser.add_argument('--clean_term_path', type=str, default="example/cleanterms/cleanterms.txt.example")
parser.add_argument('--debug', action="store_true")
parser.add_argument(
"--model_name_or_path",
default="../pretrain/pubmedbert_abs/",
type=str,
help="Path to pre-trained model or shortcut name selected in the list: ",
)
parser.add_argument(
"--output_dir",
default="",
type=str,
help="The output directory where the model predictions and checkpoints will be written.",
)
parser.add_argument(
"--save_step",
default=10000,
type=int,
help="Save step",
)
parser.add_argument(
"--eval_step",
default=10000,
type=int,
help="Save step",
)
parser.add_argument(
"--window_size",
default=32,
type=int,
help="Window size",
)
parser.add_argument("--do_train", action="store_true")
parser.add_argument(
"--train_batch_size", default=512, type=int, help="Batch size per GPU/CPU for training.",
)
parser.add_argument(
"--gradient_accumulation_steps",
type=int,
default=8,
help="Number of updates steps to accumulate before performing a backward/update pass.",
)
parser.add_argument("--learning_rate", default=1e-4,
type=float, help="The initial learning rate for Adam.")
parser.add_argument("--weight_decay", default=0.01,
type=float, help="Weight decay if we apply some.")
parser.add_argument("--adam_epsilon", default=1e-8,
type=float, help="Epsilon for Adam optimizer.")
parser.add_argument("--max_grad_norm", default=1.0,
type=float, help="Max gradient norm.")
parser.add_argument("--device", type=str, default='cuda:0', help="device")
parser.add_argument("--num_workers", default=1, type=int,
help="Num workers for data loader, only 0 can be used for Windows")
parser.add_argument("--train_epoch", default=2, type=int)
parser.add_argument("--mask_ratio", default=0.15, type=float)
args = parser.parse_args()
run(args)
if __name__ == "__main__":
main()