forked from jaeseokbyun/GRIT-VLP
-
Notifications
You must be signed in to change notification settings - Fork 0
/
Pretrain.py
243 lines (196 loc) · 9.62 KB
/
Pretrain.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
'''
* Copyright (c) 2021, salesforce.com, inc.
* All rights reserved.
* SPDX-License-Identifier: BSD-3-Clause
* For full license text, see LICENSE.txt file in the repo root or https://opensource.org/licenses/BSD-3-Clause
'''
import argparse
import os
import ruamel.yaml as yaml
import numpy as np
import random
import time
import datetime
import json
from pathlib import Path
import torch
import torch.nn as nn
import torch.nn.functional as F
from torch.utils.data import DataLoader
import torch.backends.cudnn as cudnn
import torch.distributed as dist
from models.model_pretrain_GRIT_VLP import ALBEF_Base
from models.vit import interpolate_pos_embed
from models.tokenization_bert import BertTokenizer
from models.GRIT_utils import GRIT,chunks, mini_batch_level_shuffle
import utils
from dataset.handle_data import create_dataset, create_sampler,create_fixed_sampler, create_loader
import scipy.io as sio
from scheduler import create_scheduler
from optim import create_optimizer
import os
import math
def train(model, data_loader, optimizer, tokenizer, epoch, warmup_steps, device, scheduler, config):
# train
model.train()
metric_logger = utils.MetricLogger(delimiter=" ")
metric_logger.add_meter('lr', utils.SmoothedValue(window_size=50, fmt='{value:.6f}'))
metric_logger.add_meter('loss_mlm', utils.SmoothedValue(window_size=50, fmt='{value:.4f}'))
metric_logger.add_meter('loss_ita', utils.SmoothedValue(window_size=50, fmt='{value:.4f}'))
metric_logger.add_meter('loss_itm', utils.SmoothedValue(window_size=50, fmt='{value:.4f}'))
header = 'Train Epoch: [{}]'.format(epoch)
print_freq = 50
step_size = 100
warmup_iterations = warmup_steps*step_size
# GRIT init (queue init)
num_steps=len(data_loader)
grit= GRIT (config,device,num_steps)
for i, (image, text,idx) in enumerate(metric_logger.log_every(data_loader, print_freq, header)):
optimizer.zero_grad()
idx=idx.to(device)
image = image.to(device,non_blocking=True)
text_input = tokenizer(text, padding='longest', truncation=True, max_length=25, return_tensors="pt").to(device)
# model pre-training loss
loss_mlm, loss_ita, loss_itm,image_feat_store,text_feat_store = model(image, text_input)
loss = loss_mlm + loss_ita + loss_itm
loss.backward()
optimizer.step()
# GRIT 1 phases
grit.collecting(image_feat_store,text_feat_store,idx)
# GRIT 2-3 phases
grit.grit_second_third_phase(i, model.module.temp,num_steps)
# logger
metric_logger.update(loss_mlm=loss_mlm.item())
metric_logger.update(loss_ita=loss_ita.item())
metric_logger.update(loss_itm=loss_itm.item())
metric_logger.update(lr=optimizer.param_groups[0]["lr"])
if epoch==0 and i%step_size==0 and i<=warmup_iterations:
scheduler.step(i//step_size)
dist.barrier()
# Mini-batch level shuffle (Phase 4)
if not args.mini_batch_shuffle_across_gpu:
G_index_set = mini_batch_level_shuffle(grit.G_index_set,config['batch_size'])
else:
G_index_set=grit.G_index_set
# gather the stats from all processes
metric_logger.synchronize_between_processes()
print("Averaged stats:", metric_logger.global_avg())
return {k: "{:.3f}".format(meter.global_avg) for k, meter in metric_logger.meters.items()} ,G_index_set
def main(args, config):
utils.init_distributed_mode(args)
device = torch.device(args.device)
# fix the seed
seed = args.seed + utils.get_rank()
torch.manual_seed(seed)
np.random.seed(seed)
random.seed(seed)
cudnn.benchmark = True
start_epoch = 0
max_epoch = config['train_epochs']
warmup_steps = config['schedular']['warmup_epochs']
#### Dataset ####
print("Creating dataset")
datasets = [create_dataset('pretrain', config)]
if args.distributed:
num_tasks = utils.get_world_size()
global_rank = utils.get_rank()
tokenizer = BertTokenizer.from_pretrained(args.text_encoder)
#### Model ####
print("Creating model")
model = ALBEF_Base(config=config, text_encoder=args.text_encoder, tokenizer=tokenizer, init_deit=True)
model = model.to(device)
arg_opt = utils.AttrDict(config['optimizer'])
optimizer = create_optimizer(arg_opt, model)
arg_sche = utils.AttrDict(config['schedular'])
lr_scheduler, _ = create_scheduler(arg_sche, optimizer)
if args.checkpoint:
checkpoint = torch.load(args.checkpoint, map_location='cpu')
state_dict = checkpoint['model']
if args.resume:
optimizer.load_state_dict(checkpoint['optimizer'])
lr_scheduler.load_state_dict(checkpoint['lr_scheduler'])
start_epoch = checkpoint['epoch']+1
else:
pos_embed_reshaped = interpolate_pos_embed(state_dict['visual_encoder.pos_embed'],model.visual_encoder)
state_dict['visual_encoder.pos_embed'] = pos_embed_reshaped
model.load_state_dict(state_dict)
print('load checkpoint from %s'%args.checkpoint)
model_without_ddp = model
if args.distributed:
model = torch.nn.parallel.DistributedDataParallel(model, device_ids=[args.gpu])
model_without_ddp = model.module
print("Start training")
start_time = time.time()
for epoch in range(start_epoch, max_epoch):
if epoch>0:
lr_scheduler.step(epoch+warmup_steps)
if epoch==0 and not args.index_warmstart:
samplers = create_sampler(datasets, [True], num_tasks, global_rank)
data_loader = create_loader(datasets,samplers,batch_size=[config['batch_size']], num_workers=[4], is_trains=[True], collate_fns=[None])[0]
if args.distributed:
data_loader.sampler.set_epoch(epoch)
else:
index_file= sio.loadmat(args.output_dir+'/total_indices.mat')
previous_index_set= index_file['indices'][0]
# Data loader
samplers = create_fixed_sampler(num_tasks, global_rank,previous_index_set)
data_loader = create_loader(datasets,samplers,batch_size=[config['batch_size']], num_workers=[4], is_trains=[True], collate_fns=[None])[0]
# Training
train_stats, next_index_set = train(model, data_loader, optimizer, tokenizer, epoch, warmup_steps, device, lr_scheduler, config)
# Output index save
next_index_set = torch.tensor(next_index_set,dtype=torch.long).to(device)
total_next_index_set = concat_all_gather(next_index_set)
total_next_index_set = total_next_index_set.detach().cpu().numpy()
# Mini-batch level shuffle across gpu (Phase 4)
if args.mini_batch_shuffle_across_gpu:
total_next_index_set= mini_batch_level_shuffle(list(total_next_index_set),config['batch_size'])
sio.savemat(args.output_dir+'/total_indices.mat',{'indices':total_next_index_set})
dist.barrier()
if utils.is_main_process():
log_stats = {**{f'train_{k}': v for k, v in train_stats.items()},
'epoch': epoch,
}
save_obj = {
'model': model_without_ddp.state_dict(),
'optimizer': optimizer.state_dict(),
'lr_scheduler': lr_scheduler.state_dict(),
'config': config,
'epoch': epoch,
}
torch.save(save_obj, os.path.join(args.output_dir, 'checkpoint_%02d.pth'%epoch))
with open(os.path.join(args.output_dir, "log.txt"),"a") as f:
f.write(json.dumps(log_stats) + "\n")
dist.barrier()
total_time = time.time() - start_time
total_time_str = str(datetime.timedelta(seconds=int(total_time)))
print('Training time {}'.format(total_time_str))
@torch.no_grad()
def concat_all_gather(tensor):
"""
Performs all_gather operation on the provided tensors.
*** Warning ***: torch.distributed.all_gather has no gradient.
"""
tensors_gather = [torch.ones_like(tensor)
for _ in range(torch.distributed.get_world_size())]
torch.distributed.all_gather(tensors_gather, tensor, async_op=False)
output = torch.cat(tensors_gather, dim=0)
return output
if __name__ == '__main__':
parser = argparse.ArgumentParser()
parser.add_argument('--config', default='./configs/Pretrain.yaml')
parser.add_argument('--checkpoint', default='')
parser.add_argument('--resume', default=False, type=bool)
parser.add_argument('--output_dir', default='Pretrain/')
parser.add_argument('--text_encoder', default='bert-base-uncased')
parser.add_argument('--device', default='cuda')
parser.add_argument('--seed', default=42, type=int)
parser.add_argument('--world_size', default=1, type=int, help='number of distributed processes')
parser.add_argument('--dist_url', default='env://', help='url used to set up distributed training')
parser.add_argument('--distributed', default=True, type=bool)
parser.add_argument('--index_warmstart', default=False, type=bool)
parser.add_argument('--mini_batch_shuffle_across_gpu', default=False, type=bool)
args = parser.parse_args()
config = yaml.load(open(args.config, 'r'), Loader=yaml.Loader)
Path(args.output_dir).mkdir(parents=True, exist_ok=True)
yaml.dump(config, open(os.path.join(args.output_dir, 'config.yaml'), 'w'))
main(args, config)