Repository navigation
Expand file tree
/
Copy pathmain_prvr.py
More file actions
executable file
·342 lines (290 loc) · 15.5 KB
/
Copy pathmain_prvr.py
File metadata and controls
executable file
·342 lines (290 loc) · 15.5 KB
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
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
# coding=utf-8
from __future__ import (absolute_import, division, unicode_literals)
import os
import sys
import torch
import logging
import numpy as np
import torch.nn.functional as F
import torch.multiprocessing as mp
from torch import optim, distributed
from torch.cuda.amp import GradScaler
# from torch.utils.tensorboard import SummaryWriter
from params import get_args, save_hp_to_json
from modules import CLIP4Clip_PRVR, convert_weights
from modules import SimpleTokenizer as ClipTokenizer
from modules.file import PYTORCH_PRETRAINED_BERT_CACHE
from dataloaders.dataset import get_train_dataloader, get_val_txt_dataloader, get_val_vis_dataloader
from utils.lr_scheduler import lr_scheduler
from utils.optimization import BertAdam, prep_optim_params_groups
from utils.log import set_logger
from utils.misc import set_random_seed, convert_models_to_fp32, save_checkpoint
from utils.dist_utils import is_master, get_rank, is_dist_avail_and_initialized, init_distributed_mode
from utils.metrics import compute_metrics_prvr
from tqdm import tqdm
def main(args):
"""main worker"""
args.gpu=0
args.distributed=False # we only support single gpu for now (codes for gathering text labels with dynamic lengths are so cumbersome)
set_logger(os.path.join(args.output_dir, "log.txt"), args.log_level)
set_random_seed(args.seed)
torch.backends.cudnn.benchmark = False
torch.backends.cudnn.deterministic = True
save_hp_to_json(args.output_dir, args)
tokenizer = ClipTokenizer()
model_state_dict = torch.load(args.init_model, map_location='cpu') if args.init_model else None
cache_dir = args.cache_dir if args.cache_dir else os.path.join(str(PYTORCH_PRETRAINED_BERT_CACHE), 'distributed')
model = CLIP4Clip_PRVR.from_pretrained(args.cross_model,
cache_dir=cache_dir,
state_dict=model_state_dict,
task_config=args)
if args.precision == "amp"or args.gpu is None:
logging.info("[weight convert] ==>> Convert weights to fp32 for {}...".format(args.precision))
convert_models_to_fp32(model)
logging.info("[weight convert] ==>> Convert done!")
if not torch.cuda.is_available():
model.float()
logging.warning("using CPU, this will be slow")
else:
model.cuda(args.gpu)
if args.precision == "fp16":
convert_weights(model)
if args.distributed and args.use_bn_sync:
model = torch.nn.SyncBatchNorm.convert_sync_batchnorm(model)
if args.distributed:
if args.freeze_clip:
print('do freeze_clip:',args.freeze_clip)
model = torch.nn.parallel.DistributedDataParallel(model, device_ids=[args.gpu],
find_unused_parameters=True if args.freeze_clip else True)
if args.dp:
model = torch.nn.DataParallel(model, device_ids=args.multigpu)
device = torch.device("cuda" if torch.cuda.is_available() else "cpu", 0)
## ####################################
# dataloader loading
## ####################################
val_vis_dataloader, val_vis_length = get_val_vis_dataloader(args)
val_txt_dataloader, val_txt_length = get_val_txt_dataloader(args, tokenizer)
train_dataloader, train_length, train_sampler = get_train_dataloader(args, tokenizer)
num_train_optimization_steps = (int(len(train_dataloader) + args.gradient_accumulation_steps - 1)
/ args.gradient_accumulation_steps) * args.epochs
## ####################################
# optimization strategies
## ####################################
optimizer_grouped_parameters = prep_optim_params_groups(args, model, coef_lr=args.coef_lr)
scaler = GradScaler() if args.precision == "amp" else None
if args.optim == 'BertAdam':
logging.info('[optimizer] Using BertAdam Optimizer...')
optimizer = BertAdam(optimizer_grouped_parameters, lr=args.lr, warmup=args.warmup_proportion,
schedule='warmup_cosine', b1=0.9, b2=0.98, e=1e-6,
t_total=num_train_optimization_steps, weight_decay=args.wd,
max_grad_norm=1.0)
scheduler = None
elif args.optim == 'AdamW':
logging.info('[optimizer] Using AdamW Optimizer...')
optimizer = optim.AdamW(optimizer_grouped_parameters, lr=args.lr,
betas=(args.beta1, args.beta2), eps=args.eps, weight_decay=args.wd)
scheduler = lr_scheduler(mode='cos', init_lr=args.lr, all_iters=num_train_optimization_steps,
slow_start_iters=args.warmup_proportion * num_train_optimization_steps,
weight_decay=args.wd
)
else:
raise NotImplementedError
## ####################################
# optionally resume from a checkpoint
## ####################################
start_epoch, global_step = 0, 0
if args.resume is not None:
if os.path.isfile(args.resume):
if args.gpu is None:
checkpoint = torch.load(args.resume)
else:
loc = "cuda:{}".format(args.gpu)
checkpoint = torch.load(args.resume, map_location=loc)
sd = checkpoint["state_dict"]
if not args.distributed and next(iter(sd.items()))[0].startswith('module'):
sd = {k[len('module.'):]: v for k, v in sd.items()}
model.load_state_dict(sd)
if not args.load_from_pretrained:
if "optimizer" in checkpoint and optimizer is not None:
optimizer.load_state_dict(checkpoint["optimizer"])
if "scaler" in checkpoint and scaler is not None:
logging.info("[resume] => Loading state_dict of AMP loss scaler")
scaler.load_state_dict(checkpoint['scaler'])
start_epoch, global_step = checkpoint["epoch"], checkpoint["global_step"]
logging.info(f"[resume] => loaded checkpoint '{args.resume}' (epoch {checkpoint['epoch']})\n")
else:
logging.info("[resume] => no checkpoint found at '{}'\n".format(args.resume))
## ####################################
# train and evalution
## ####################################
logging.info("\n======================== Running training ========================")
logging.info(" Num examples = %d", train_length)
logging.info(" Batch size = %d", args.batch_size)
logging.info(" Num steps = %d", num_train_optimization_steps * args.gradient_accumulation_steps)
logging.info("\n======================== Running val ========================")
logging.info(" Num videos = %d", val_vis_length)
logging.info(" Num text = %d", val_txt_length)
logging.info(" Batch size = %d", args.batch_size_val)
logging.info(" Num steps = %d", len(val_txt_dataloader))
if args.do_eval:
tv_metrics, info_str = prvr_eval_epoch(model, val_vis_dataloader, val_txt_dataloader, device, args=args)
torch.cuda.synchronize()
sys.exit(0)
best_R, best_e = 0, 0
best_info = []
stop_R = 'SumR'
for epoch in range(start_epoch, args.epochs):
tr_loss, global_step = train_epoch(epoch, args, model, train_dataloader, device, optimizer,
global_step, scaler=scaler, scheduler=scheduler)
logging.info("Epoch %d/%s Finished, Train Loss: %f", epoch + 1, args.epochs, tr_loss)
tv_metrics, info_str = prvr_eval_epoch(model, val_vis_dataloader, val_txt_dataloader, device, args=args)
if best_R <= tv_metrics[stop_R]:
best_R = tv_metrics[stop_R]
best_e = epoch
best_info = info_str
logging.info('Got New Best Model')
else:
logging.info('A relative failed epoch')
logging.info("The best {} is: {:.4f}, best_e={}\n".format(stop_R, best_R, best_e))
ckpt_dict = {
'epoch': epoch + 1,
'global_step': global_step,
'arch': 'CLIp4Clip',
'state_dict': model.state_dict(),
'best_acc1': best_R,
'optimizer': optimizer.state_dict(),
}
if scaler is not None: ckpt_dict['scaler'] = scaler.state_dict()
save_checkpoint(ckpt_dict, best_R <= tv_metrics[stop_R], args.output_dir, filename='ckpt.pth.tar')
logging.info("The best {} is: {:.4f}, best_epoch={}\n".format(stop_R, best_R, best_e))
for info in best_info:
logging.info(info)
print("The above program id is {}\n".format(args.output_dir))
torch.cuda.empty_cache()
sys.exit(0)
def train_epoch(epoch, args, model, train_dataloader, device, optimizer, global_step,
scheduler=None, scaler=None):
samples_per_epoch = len(train_dataloader.dataset)
model.train()
if epoch == 0:
trainable_size =0
total_param_size = 0
for name, param in model.named_parameters():
if param.requires_grad==True:
total_param_size += param.numel()
trainable_size += param.numel()
param_size_MB = param.numel()/(1000**2)
logging.info(f'trainerble parameters are: {name}, size is {param_size_MB:.4f} MB')
else:
total_param_size += param.numel()
trainable_size_MB = trainable_size/(1000**2)
total_param_size_MB = total_param_size/(1000**2)
percentage = (trainable_size / total_param_size)*100
logging.info("Trainable param percentage are: {}".format(percentage))
logging.info("Trainable params are: {} MB, Total params are: {} MB".format(trainable_size_MB,total_param_size_MB))
total_loss = 0
local_rank = get_rank()
for step, batch in tqdm(enumerate(train_dataloader), total=len(train_dataloader), disable=(local_rank!=0)):
optimizer.zero_grad()
if scheduler is not None: scheduler(optimizer, global_step=global_step)
if torch.cuda.is_available():
for k, v in batch.items():
if type(v) is torch.Tensor:
batch[k] = v.to(device=device, non_blocking=True)
with torch.cuda.amp.autocast(enabled=scaler is not None):
output = model(batch, epoch_id=epoch)
loss = output['loss'].mean()
sim_loss = output['sim_loss'].mean()
if args.gradient_accumulation_steps > 1:
loss = loss / args.gradient_accumulation_steps
if scaler is not None:
scaler.scale(loss).backward()
if (step + 1) % args.gradient_accumulation_steps == 0:
if args.clip_grad_norm is not None:
scaler.unscale_(optimizer)
torch.nn.utils.clip_grad_norm_(model.parameters(), args.clip_grad_norm)
scaler.step(optimizer)
scaler.update()
else:
loss.backward()
if (step + 1) % args.gradient_accumulation_steps == 0:
if args.clip_grad_norm is not None:
torch.nn.utils.clip_grad_norm_(model.parameters(), args.clip_grad_norm)
optimizer.step()
if hasattr(model, 'module'):
torch.clamp_(model.module.clip.logit_scale.data, 0.1, 4.6052)
else:
torch.clamp_(model.clip.logit_scale.data, 0.1, 4.6052)
if (step + 1) % args.gradient_accumulation_steps == 0:
global_step += 1
if global_step % args.n_display == 0:
num_samples = (step + 1) * len(batch['video_name']) * args.world_size
percent_complete = num_samples * 1.0 / samples_per_epoch * 100
logit_scale_data = model.module.clip.logit_scale.data if hasattr(model, 'module') \
else model.clip.logit_scale.data
lr_tmp = optimizer.param_groups[0]['lr'] if args.optim == 'AdamW' else \
optimizer.get_lr()[0]
logging.info(
f"Epoch: {epoch} [{num_samples} ({percent_complete:.1f}%)]\t"
f"SimLoss: {sim_loss.item():.4f} \t"
f"\tLR: {lr_tmp:.1e}\tlogit_scale {logit_scale_data:.3f}"
)
total_loss += float(loss)
total_loss = total_loss / len(train_dataloader)
return total_loss, global_step
def prvr_eval_epoch(model, val_vis_dataloader, val_txt_dataloader, device, args=None):
model.eval()
local_rank = get_rank()
GT_TXT2VID = val_txt_dataloader.dataset.txt2vid
torch.cuda.empty_cache()
with torch.no_grad():
text_names, video_names = [], []
batch_sequence_output_list, batch_visual_output_list = [], []
logging.info("calculating text features...")
for bid, batch in tqdm(enumerate(val_txt_dataloader), total=len(val_txt_dataloader), disable=(local_rank!=0)):
for k, v in batch.items():
if type(v) is torch.Tensor:
batch[k] = v.to(device=device, non_blocking=True)
sequence_output = model.encode_text(batch['text_ids'])
batch_sequence_output_list.append(sequence_output.cpu())
torch.cuda.empty_cache()
text_names += batch['text_name']
logging.info("calculating video features...")
for bid, batch in tqdm(enumerate(val_vis_dataloader), total=len(val_vis_dataloader), disable=(local_rank!=0)):
batch = tuple(t.to(device=device, non_blocking=True) if type(t) is not tuple else t for t in batch)
video_name, video = batch
visual_output = model.encode_video(video)
batch_visual_output_list.append(visual_output.cpu())
torch.cuda.empty_cache()
video_names += video_name
if torch.cuda.is_available(): torch.cuda.synchronize()
sim_matrix = _prvr_run_on_single_gpu(model, batch_sequence_output_list, batch_visual_output_list, args=args)
logging.info("sim matrix size: {}, {}".format(sim_matrix.shape[0], sim_matrix.shape[1]))
tv_metrics = compute_metrics_prvr(sim_matrix, text_names, video_names, gt_txt2vid=GT_TXT2VID)
logging.info('\t Length-T: {}, Length-V:{}'.format(len(sim_matrix), len(sim_matrix[0])))
info_str = []
info_str.append("Text-to-Video:")
info_str.append(' (metric) >>> R@1: {:.1f} - R@5: {:.1f} - R@10: {:.1f} - R@100: {:.1f} - SumR: {:.1f} - Median R: {:.1f} - Mean R: {:.1f}'.
format(tv_metrics['R1'], tv_metrics['R5'], tv_metrics['R10'], tv_metrics['R100'], tv_metrics['SumR'], tv_metrics['MR'], tv_metrics['MeanR']))
for info in info_str: logging.info(info)
return tv_metrics, info_str
def _prvr_run_on_single_gpu(model, batch_sequence_output_list, batch_visual_output_list, args=None):
if hasattr(model, 'module'):
model = model.module
else:
model = model
sim_matrix = []
for sequence_output in batch_sequence_output_list:
each_row = []
for visual_output in batch_visual_output_list:
b1b2_logits = model.get_prvr_similarity_logits(sequence_output, visual_output)
torch.cuda.empty_cache()
b1b2_logits = b1b2_logits.cpu().detach().numpy()
each_row.append(b1b2_logits)
each_row = np.concatenate(tuple(each_row), axis=-1)
sim_matrix.append(each_row)
sim_matrix = np.concatenate(tuple(sim_matrix), axis=0)
return sim_matrix
if __name__ == "__main__":
args = get_args()
main(args)