def _gradient_accumulation(self, true_batchs, normalization, total_stats,
report_stats):
if self.grad_accum_count > 1:
self.model.zero_grad()
for batch in true_batchs:
if self.grad_accum_count == 1:
self.model.zero_grad()
src = batch.src
labels = batch.labels
segs = batch.segs
clss = batch.clss
mask = batch.mask
mask_cls = batch.mask_cls
sent_scores, mask = self.model(src, segs, clss, mask, mask_cls)
loss = self.loss(sent_scores, labels.float())
loss = (loss * mask.float()).sum()
(loss / loss.numel()).backward()
# loss.div(float(normalization)).backward()
batch_stats = Statistics(float(loss.cpu().data.numpy()),
normalization)
total_stats.update(batch_stats)
report_stats.update(batch_stats)
# 4. Update the parameters and statistics.
if self.grad_accum_count == 1:
# Multi GPU gradient gather
if self.n_gpu > 1:
grads = [
p.grad.data for p in self.model.parameters()
if p.requires_grad and p.grad is not None
]
distributed.all_reduce_and_rescale_tensors(grads, float(1))
self.optimizer.step()
# in case of multi step gradient accumulation,
# update only after accum batches
if self.grad_accum_count > 1:
if self.n_gpu > 1:
grads = [
p.grad.data for p in self.model.parameters()
if p.requires_grad and p.grad is not None
]
distributed.all_reduce_and_rescale_tensors(grads, float(1))
self.optimizer.step()
def main(opts):
hvd.init()
n_gpu = hvd.size()
device = torch.device("cuda", hvd.local_rank())
torch.cuda.set_device(hvd.local_rank())
rank = hvd.rank()
opts.rank = rank
LOGGER.info("device: {} n_gpu: {}, rank: {}, "
"16-bits training: {}".format(device, n_gpu, hvd.rank(),
opts.fp16))
if opts.gradient_accumulation_steps < 1:
raise ValueError("Invalid gradient_accumulation_steps parameter: {}, "
"should be >= 1".format(
opts.gradient_accumulation_steps))
set_random_seed(opts.seed)
if hvd.rank() == 0:
save_training_meta(opts)
TB_LOGGER.create(join(opts.output_dir, 'log'))
pbar = tqdm(total=opts.num_train_steps)
model_saver = ModelSaver(join(opts.output_dir, 'ckpt'))
add_log_to_file(join(opts.output_dir, 'log', 'log.txt'))
# store ITM predictions
os.makedirs(join(opts.output_dir, 'results_val'))
os.makedirs(join(opts.output_dir, 'results_test'))
os.makedirs(join(opts.output_dir, 'results_train'))
else:
LOGGER.disabled = True
pbar = NoOp()
model_saver = NoOp()
# train_examples = None
LOGGER.info(f"Loading Train Dataset {opts.train_txt_dbs}, "
f"{opts.train_img_dbs}")
# check multiple DBs
assert len(opts.train_txt_dbs) == len(opts.train_img_dbs), \
"train txt_db and img_db have different length"
# load DBs and image dirs
all_img_dbs = ImageLmdbGroup(opts.conf_th, opts.max_bb, opts.min_bb,
opts.num_bb, opts.compressed_db)
# train
LOGGER.info(f"Loading Train Dataset "
f"{opts.train_txt_dbs}, {opts.train_img_dbs}")
train_datasets = []
for txt_path, img_path in zip(opts.train_txt_dbs, opts.train_img_dbs):
img_db = all_img_dbs[img_path]
txt_db = TxtTokLmdb(txt_path, opts.max_txt_len)
train_datasets.append(
ItmRankDataset(txt_db, img_db, opts.negative_size))
train_dataset = ConcatDataset(train_datasets)
# val
LOGGER.info(f"Loading Val Dataset {opts.val_txt_db}, {opts.val_img_db}")
val_img_db = all_img_dbs[opts.val_img_db]
val_txt_db = TxtTokLmdb(opts.val_txt_db, -1)
val_dataset = ItmValDataset(val_txt_db, val_img_db,
opts.inf_minibatch_size)
val_dataloader = build_dataloader(val_dataset, itm_val_collate, False,
opts)
# eval
LOGGER.info(f"Loading val, test Dataset for full evaluation: "
f"{opts.val_txt_db}, {opts.val_img_db}"
f"{opts.test_txt_db}, {opts.test_img_db}")
eval_dataset_val = ItmEvalDataset(val_txt_db, val_img_db,
opts.inf_minibatch_size)
eval_loader_val = build_dataloader(eval_dataset_val, itm_eval_collate,
False, opts)
test_img_db = all_img_dbs[opts.test_img_db]
test_txt_db = TxtTokLmdb(opts.test_txt_db, -1)
eval_dataset_test = ItmEvalDataset(test_txt_db, test_img_db,
opts.inf_minibatch_size)
eval_loader_test = build_dataloader(eval_dataset_test, itm_eval_collate,
False, opts)
# Prepare model
if opts.checkpoint:
checkpoint = torch.load(opts.checkpoint)
else:
checkpoint = {}
model = UniterForImageTextRetrieval.from_pretrained(opts.model_config,
state_dict=checkpoint,
img_dim=IMG_DIM,
margin=opts.margin)
model.init_output() # pretrain ITM head is different from ranking head
model.to(device)
# make sure every process has same model parameters in the beginning
broadcast_tensors([p.data for p in model.parameters()], 0)
set_dropout(model, opts.dropout)
# Prepare optimizer
optimizer = build_optimizer(model, opts)
model, optimizer = amp.initialize(model,
optimizer,
enabled=opts.fp16,
opt_level='O2')
global_step = 0
LOGGER.info(f"***** Running training on {n_gpu} GPUs *****")
LOGGER.info(" Num examples = %d", len(train_dataset) * hvd.size())
LOGGER.info(" Batch size = %d", opts.train_batch_size)
LOGGER.info(" Accumulate steps = %d", opts.gradient_accumulation_steps)
LOGGER.info(" Num steps = %d", opts.num_train_steps)
running_loss = RunningMeter('loss')
model.train()
n_examples = 0
n_epoch = 0
start = time()
# quick hack for amp delay_unscale bug
optimizer.zero_grad()
optimizer.step()
while True:
train_dataloader = build_dataloader(train_dataset, itm_rank_collate,
True, opts)
for step, batch in enumerate(train_dataloader):
n_examples += batch['input_ids'].size(0)
loss = model(batch, compute_loss=True)
loss = loss.mean()
delay_unscale = (step + 1) % opts.gradient_accumulation_steps != 0
with amp.scale_loss(loss, optimizer,
delay_unscale=delay_unscale) as scaled_loss:
scaled_loss.backward()
if not delay_unscale:
# gather gradients from every processes
# do this before unscaling to make sure every process uses
# the same gradient scale
grads = [
p.grad.data for p in model.parameters()
if p.requires_grad and p.grad is not None
]
all_reduce_and_rescale_tensors(grads, float(1))
running_loss(loss.item())
if (step + 1) % opts.gradient_accumulation_steps == 0:
global_step += 1
# learning rate scheduling
lr_this_step = get_lr_sched(global_step, opts)
for param_group in optimizer.param_groups:
param_group['lr'] = lr_this_step
TB_LOGGER.add_scalar('lr', lr_this_step, global_step)
# log loss
# NOTE: not gathered across GPUs for efficiency
TB_LOGGER.add_scalar('loss', running_loss.val, global_step)
TB_LOGGER.step()
# update model params
if opts.grad_norm != -1:
grad_norm = clip_grad_norm_(amp.master_params(optimizer),
opts.grad_norm)
TB_LOGGER.add_scalar('grad_norm', grad_norm, global_step)
optimizer.step()
optimizer.zero_grad()
pbar.update(1)
if global_step % 100 == 0:
# monitor training throughput
LOGGER.info(f'------------Step {global_step}-------------')
tot_ex = sum(all_gather_list(n_examples))
ex_per_sec = int(tot_ex / (time() - start))
LOGGER.info(f'{tot_ex} examples trained at '
f'{ex_per_sec} ex/s')
TB_LOGGER.add_scalar('perf/ex_per_s', ex_per_sec,
global_step)
LOGGER.info(f'-------------------------------------------')
if global_step % opts.valid_steps == 0:
if opts.full_val:
LOGGER.info(
f"========================== Step {global_step} "
f"==========================")
val_log = evaluate(model, eval_loader_val)
TB_LOGGER.log_scaler_dict(
{f"valid/{k}": v
for k, v in val_log.items()})
LOGGER.info(f"image retrieval R1: "
f"{val_log['img_r1']*100:.2f},\n"
f"image retrieval R5: "
f"{val_log['img_r5']*100:.2f},\n"
f"image retrieval R10: "
f"{val_log['img_r10']*100:.2f}\n"
f"text retrieval R1: "
f"{val_log['txt_r1']*100:.2f},\n"
f"text retrieval R5: "
f"{val_log['txt_r5']*100:.2f},\n"
f"text retrieval R10: "
f"{val_log['txt_r10']*100:.2f}")
LOGGER.info("================================="
"=================================")
else:
val_log = validate(model, val_dataloader)
TB_LOGGER.log_scaler_dict(val_log)
model_saver.save(model, global_step)
if global_step >= opts.num_train_steps:
break
if global_step >= opts.num_train_steps:
break
n_epoch += 1
LOGGER.info(f"finished {n_epoch} epochs")
pbar.close()
if opts.num_train_steps % opts.valid_steps != 0:
# final validation
val_log = validate(model, val_dataloader)
TB_LOGGER.log_scaler_dict(val_log)
model_saver.save(model, global_step)
# evaluation
for split, loader in [('val', eval_loader_val),
('test', eval_loader_test)]:
eval_log = evaluate(model, loader)
TB_LOGGER.log_scaler_dict(
{f"eval/{split}_{k}": v
for k, v in eval_log.items()})
if hvd.rank() != 0:
continue
LOGGER.info(
f"========================= {split} ===========================\n"
f"image retrieval R1: {eval_log['img_r1']*100:.2f},\n"
f"image retrieval R5: {eval_log['img_r5']*100:.2f},\n"
f"image retrieval R10: {eval_log['img_r10']*100:.2f}\n"
f"text retrieval R1: {eval_log['txt_r1']*100:.2f},\n"
f"text retrieval R5: {eval_log['txt_r5']*100:.2f},\n"
f"text retrieval R10: {eval_log['txt_r10']*100:.2f}")
LOGGER.info("=========================================================")
def main(opts):
hvd.init()
n_gpu = hvd.size()
device = torch.device("cuda", hvd.local_rank())
torch.cuda.set_device(hvd.local_rank())
rank = hvd.rank()
opts.rank = rank
LOGGER.info("device: {} n_gpu: {}, rank: {}, "
"16-bits training: {}".format(device, n_gpu, hvd.rank(),
opts.fp16))
if opts.gradient_accumulation_steps < 1:
raise ValueError("Invalid gradient_accumulation_steps parameter: {}, "
"should be >= 1".format(
opts.gradient_accumulation_steps))
set_random_seed(opts.seed)
# train_examples = None
LOGGER.info(f"Loading Train Dataset {opts.train_txt_db}, "
f"{opts.train_img_db}")
train_dataloader = create_dataloader(opts.train_img_db, opts.train_txt_db,
opts.train_batch_size, True,
ReDataset, re_collate, opts)
val_dataloader = create_dataloader(opts.val_img_db, opts.val_txt_db,
opts.val_batch_size, False,
ReEvalDataset, re_eval_collate, opts)
# Prepare model
if opts.checkpoint:
checkpoint = torch.load(opts.checkpoint)
else:
checkpoint = {}
all_dbs = [opts.train_txt_db, opts.val_txt_db]
toker = json.load(open(f'{all_dbs[0]}/meta.json'))['toker']
assert all(toker == json.load(open(f'{db}/meta.json'))['toker']
for db in all_dbs)
model = UniterForReferringExpressionComprehension.from_pretrained(
opts.model_config,
checkpoint,
img_dim=IMG_DIM,
loss=opts.train_loss,
margin=opts.margin,
hard_ratio=opts.hard_ratio,
mlp=opts.mlp,
)
model.to(device)
model.train()
# make sure every process has same model parameters in the beginning
broadcast_tensors([p.data for p in model.parameters()], 0)
set_dropout(model, opts.dropout)
optimizer = build_optimizer(model, opts)
# Apex
model, optimizer = amp.initialize(model,
optimizer,
enabled=opts.fp16,
opt_level='O2')
global_step = 0
if rank == 0:
save_training_meta(opts)
TB_LOGGER.create(join(opts.output_dir, 'log'))
pbar = tqdm(total=opts.num_train_steps)
model_saver = ModelSaver(join(opts.output_dir, 'ckpt'), 'model_epoch')
os.makedirs(join(opts.output_dir, 'results')) # store RE predictions
add_log_to_file(join(opts.output_dir, 'log', 'log.txt'))
else:
LOGGER.disabled = True
pbar = NoOp()
model_saver = NoOp()
LOGGER.info(f"***** Running training with {n_gpu} GPUs *****")
LOGGER.info(" Num examples = %d", len(train_dataloader.dataset))
LOGGER.info(" Batch size = %d", opts.train_batch_size)
LOGGER.info(" Accumulate steps = %d", opts.gradient_accumulation_steps)
LOGGER.info(" Num steps = %d", opts.num_train_steps)
running_loss = RunningMeter('loss')
model.train()
n_examples = 0
n_epoch = 0
best_val_acc, best_epoch = None, None
start = time()
# quick hack for amp delay_unscale bug
optimizer.zero_grad()
if global_step == 0:
optimizer.step()
while True:
for step, batch in enumerate(train_dataloader):
if global_step >= opts.num_train_steps:
break
n_examples += batch['input_ids'].size(0)
loss = model(batch, compute_loss=True)
loss = loss.sum() # sum over vectorized loss TODO: investigate
delay_unscale = (step + 1) % opts.gradient_accumulation_steps != 0
with amp.scale_loss(loss, optimizer,
delay_unscale=delay_unscale) as scaled_loss:
scaled_loss.backward()
if not delay_unscale:
# gather gradients from every processes
# do this before unscaling to make sure every process uses
# the same gradient scale
grads = [
p.grad.data for p in model.parameters()
if p.requires_grad and p.grad is not None
]
all_reduce_and_rescale_tensors(grads, float(1))
running_loss(loss.item())
if (step + 1) % opts.gradient_accumulation_steps == 0:
global_step += 1
# learning rate scheduling
lr_this_step = get_lr_sched(global_step, opts)
for i, param_group in enumerate(optimizer.param_groups):
if i == 0 or i == 1:
param_group['lr'] = lr_this_step * opts.lr_mul
elif i == 2 or i == 3:
param_group['lr'] = lr_this_step
else:
raise ValueError()
TB_LOGGER.add_scalar('lr', lr_this_step, global_step)
# log loss
# NOTE: not gathered across GPUs for efficiency
TB_LOGGER.add_scalar('loss', running_loss.val, global_step)
TB_LOGGER.step()
# update model params
if opts.grad_norm != -1:
grad_norm = clip_grad_norm_(amp.master_params(optimizer),
opts.grad_norm)
TB_LOGGER.add_scalar('grad_norm', grad_norm, global_step)
optimizer.step()
optimizer.zero_grad()
pbar.update(1)
if global_step % 100 == 0:
# monitor training throughput
LOGGER.info(f'============Step {global_step}=============')
tot_ex = sum(all_gather_list(n_examples))
ex_per_sec = int(tot_ex / (time() - start))
LOGGER.info(f'{tot_ex} examples trained at '
f'{ex_per_sec} ex/s')
TB_LOGGER.add_scalar('perf/ex_per_s', ex_per_sec,
global_step)
LOGGER.info('===========================================')
# evaluate after each epoch
val_log, _ = validate(model, val_dataloader)
TB_LOGGER.log_scaler_dict(val_log)
# save model
n_epoch += 1
model_saver.save(model, n_epoch)
LOGGER.info(f"finished {n_epoch} epochs")
# save best model
if best_val_acc is None or val_log['valid/acc'] > best_val_acc:
best_val_acc = val_log['valid/acc']
best_epoch = n_epoch
model_saver.save(model, 'best')
# shuffle training data for the next epoch
train_dataloader.loader.dataset.shuffle()
# is training finished?
if global_step >= opts.num_train_steps:
break
val_log, results = validate(model, val_dataloader)
with open(
f'{opts.output_dir}/results/'
f'results_{global_step}_'
f'rank{rank}_final.json', 'w') as f:
json.dump(results, f)
TB_LOGGER.log_scaler_dict(val_log)
model_saver.save(model, f'{global_step}_final')
# print best model
LOGGER.info(
f'best_val_acc = {best_val_acc*100:.2f}% at epoch {best_epoch}.')
def _gradient_accumulation(self, true_batchs, normalization, total_stats,
report_stats, ratio):
if self.grad_accum_count > 1:
self.model.zero_grad()
for batch in true_batchs:
target_size = batch.tgt.size(0)
# Truncated BPTT: reminder not compatible with accum > 1
if self.trunc_size:
trunc_size = self.trunc_size
else:
trunc_size = target_size
# dec_state = None
src = make_features(batch, 'src')
_, src_lengths = batch.src
tgt_outer = make_features(batch, 'tgt')
structure = make_features(batch, 'structure')
structure = structure.transpose(0, 1)
structure = structure.transpose(1, 2)
# bad code
mask = make_features(batch, 'mask')
mask = mask - 2
mask[mask <= 0] = 0
mask = mask.byte()
# ground truth label of biaffine relation
relation = make_features(batch, 'relation')
relation = relation.transpose(0, 1)
relation = relation[relation != 1]
for j in range(0, target_size - 1, trunc_size):
# 1. Create truncated target.
tgt = tgt_outer[j: j + trunc_size]
# 2. F-prop all but generator.
if self.grad_accum_count == 1:
self.model.zero_grad()
outputs, attns, p, rels = \
self.model(src, tgt, structure, mask, src_lengths)
# 3. Compute loss in shards for memory efficiency.
batch_stats = self.train_loss.sharded_compute_loss(
batch, outputs, attns, j,
trunc_size, self.shard_size, normalization, 1.)
if relation.size(0)>0:
# compute loss for label prediction
relation_loss = self.train_relation_loss(rels, relation)
# total loss of biaffine module
loss = (-p + relation_loss) / relation.size(0)
loss = loss * ratio
loss.backward()
total_stats.update(batch_stats)
report_stats.update(batch_stats)
# 4. Update the parameters and statistics.
if self.grad_accum_count == 1:
# Multi GPU gradient gather
if self.n_gpu > 1:
grads = [p.grad.data for p in self.model.parameters()
if p.requires_grad
and p.grad is not None]
all_reduce_and_rescale_tensors(
grads, float(1))
self.optim.step()
# If truncated, don't backprop fully.
# TO CHECK
# if dec_state is not None:
# dec_state.detach()
if self.model.decoder.state is not None:
self.model.decoder.detach_state()
# in case of multi step gradient accumulation,
# update only after accum batches
if self.grad_accum_count > 1:
if self.n_gpu > 1:
grads = [p.grad.data for p in self.model.parameters()
if p.requires_grad
and p.grad is not None]
all_reduce_and_rescale_tensors(
grads, float(1))
self.optim.step()
def main(opts):
hvd.init()
n_gpu = hvd.size()
device = torch.device("cuda", hvd.local_rank())
torch.cuda.set_device(hvd.local_rank())
rank = hvd.rank()
opts.rank = rank
LOGGER.info("device: {} n_gpu: {}, rank: {}, "
"16-bits training: {}".format(device, n_gpu, hvd.rank(),
opts.fp16))
if hvd.rank() != 0:
LOGGER.disabled = True
if opts.gradient_accumulation_steps < 1:
raise ValueError("Invalid gradient_accumulation_steps parameter: {}, "
"should be >= 1".format(
opts.gradient_accumulation_steps))
set_random_seed(opts.seed)
opts.task = 'tvc'
# train_examples = None
LOGGER.info(f"Loading the whole video dataset {opts.sub_txt_db}, "
f"{opts.vfeat_db}")
video_db = load_video_sub_dataset(opts.vfeat_db, opts.sub_txt_db,
opts.vfeat_interval, opts)
# data loaders
# train
LOGGER.info(f"Loading train dataset {opts.train_db}")
train_cap = CaptionTokLmdb(opts.train_db, opts.max_txt_len)
train_dset = TvcTrainDataset(video_db, train_cap, opts.max_cap_per_vid)
LOGGER.info(f"{sum(all_gather_list(len(train_dset)))} samples loaded")
train_loader = build_dataloader(train_dset, opts.train_batch_size,
TvcTrainDataset.collate, True, opts)
# val
LOGGER.info(f"Loading val dataset {opts.val_db}")
val_cap = CaptionTokLmdb(opts.val_db, -1)
val_dset = TvcValDataset(video_db, val_cap, -1)
val_loader = build_dataloader(val_dset, opts.val_batch_size,
TvcValDataset.collate, False, opts)
if hvd.rank() == 0:
evaluator = TVCEval(opts.val_ref)
else:
evaluator = NoOp()
# Prepare model
if opts.checkpoint:
checkpoint = torch.load(opts.checkpoint)
else:
checkpoint = {}
img_pos_embed_weight_key = "v_encoder.f_encoder.img_embeddings" +\
".position_embeddings.weight"
if img_pos_embed_weight_key in checkpoint:
max_frm_seq_len = len(checkpoint[img_pos_embed_weight_key])
else:
max_frm_seq_len = MAX_FRM_SEQ_LEN
model = HeroForTvc.from_pretrained(opts.model_config,
state_dict=checkpoint,
vfeat_dim=VFEAT_DIM,
max_frm_seq_len=max_frm_seq_len,
lsr=opts.lsr)
model.to(device)
# make sure every process has same model parameters in the beginning
broadcast_tensors([p.data for p in model.parameters()], 0)
set_dropout(model, opts.dropout)
# Prepare optimizer
optimizer = build_optimizer(model, opts)
model, optimizer = amp.initialize(model,
optimizer,
enabled=opts.fp16,
opt_level='O2')
# assumes roberta tokenizer only
if hvd.local_rank() == 0:
# quick hack to prevent multi-process download collision
toker = RobertaTokenizer.from_pretrained('roberta-base')
all_gather_list(None)
else:
all_gather_list(None)
toker = RobertaTokenizer.from_pretrained('roberta-base')
bos = toker.convert_tokens_to_ids(['<s>'])[0]
eos = toker.convert_tokens_to_ids(['</s>'])[0]
generator = TvcGenerator(model, opts.max_gen_step, bos, eos, opts.fp16)
global_step = 0
if rank == 0:
save_training_meta(opts)
TB_LOGGER.create(join(opts.output_dir, 'log'))
pbar = tqdm(total=opts.num_train_steps)
model_saver = ModelSaver(join(opts.output_dir, 'ckpt'))
os.makedirs(join(opts.output_dir, 'results')) # store val predictions
add_log_to_file(join(opts.output_dir, 'log', 'log.txt'))
else:
LOGGER.disabled = True
pbar = NoOp()
model_saver = NoOp()
LOGGER.info(f"***** Running training with {n_gpu} GPUs *****")
LOGGER.info(" Batch size = %d", opts.train_batch_size)
LOGGER.info(" Accumulate steps = %d", opts.gradient_accumulation_steps)
LOGGER.info(" Num steps = %d", opts.num_train_steps)
train_loss = RunningMeter('loss')
n_vid = 0
n_cap = 0
n_epoch = 0
start = time()
# quick hack for amp delay_unscale bug
optimizer.zero_grad()
optimizer.step()
model.train()
while True:
for step, batch in enumerate(train_loader):
n_vid += opts.train_batch_size
n_cap += batch['cap_input_ids'].size(0)
loss = model(batch, compute_loss=True)
loss = loss.mean()
train_loss(loss.item())
delay_unscale = (step + 1) % opts.gradient_accumulation_steps != 0
with amp.scale_loss(loss, optimizer,
delay_unscale=delay_unscale) as scaled_loss:
scaled_loss.backward()
if not delay_unscale:
# gather gradients from every processes
# do this before unscaling to make sure every process uses
# the same gradient scale
grads = [
p.grad.data for p in model.parameters()
if p.requires_grad and p.grad is not None
]
all_reduce_and_rescale_tensors(grads, float(1))
if (step + 1) % opts.gradient_accumulation_steps == 0:
global_step += 1
# learning rate scheduling
lr_this_step = get_lr_sched(global_step, opts)
for i, param_group in enumerate(optimizer.param_groups):
if i == 0 or i == 1:
param_group['lr'] = lr_this_step * opts.lr_mul
elif i == 2 or i == 3:
param_group['lr'] = lr_this_step
else:
raise ValueError()
TB_LOGGER.add_scalar('lr', lr_this_step, global_step)
# log loss
TB_LOGGER.add_scalar(train_loss.name, train_loss.val,
global_step)
TB_LOGGER.step()
# update model params
if opts.grad_norm != -1:
grad_norm = clip_grad_norm_(amp.master_params(optimizer),
opts.grad_norm)
TB_LOGGER.add_scalar('grad_norm', grad_norm, global_step)
optimizer.step()
optimizer.zero_grad()
pbar.update(1)
if global_step % 100 == 0:
# monitor training throughput
LOGGER.info('-------------------------------------------')
LOGGER.info(f'Step {global_step}:')
tot_vid = sum(all_gather_list(n_vid))
vid_per_sec = int(tot_vid / (time() - start))
LOGGER.info(f'{tot_vid} videos trained at '
f'{vid_per_sec} vid/s')
tot_cap = sum(all_gather_list(n_cap))
cap_per_sec = int(tot_cap / (time() - start))
TB_LOGGER.add_scalar(f'perf/vid_per_s', vid_per_sec,
global_step)
TB_LOGGER.add_scalar(f'perf/cap_per_s', cap_per_sec,
global_step)
if global_step % opts.valid_steps == 0:
LOGGER.info('===========================================')
LOGGER.info(f"Step {global_step}: start validation")
val_log, results = validate(val_loader, generator, toker,
evaluator)
if hvd.rank() == 0:
save_jsonl(
results, f"{opts.output_dir}/results/"
f"/results_{global_step}.jsonl")
TB_LOGGER.log_scaler_dict(val_log)
LOGGER.info('===========================================')
model_saver.save(model, global_step)
if global_step >= opts.num_train_steps:
break
n_epoch += 1
LOGGER.info(f"finished {n_epoch} epochs")
if global_step >= opts.num_train_steps:
break
LOGGER.info('===========================================')
if global_step % opts.valid_steps != 0:
val_log, results = validate(val_loader, generator, toker, evaluator)
if hvd.rank() == 0:
save_jsonl(
results, f"{opts.output_dir}/results/"
f"/results_{global_step}.jsonl")
TB_LOGGER.log_scaler_dict(val_log)
model_saver.save(model, global_step)
def main(opts):
hvd.init()
n_gpu = hvd.size()
device = torch.device("cuda", hvd.local_rank())
torch.cuda.set_device(hvd.local_rank())
opts.n_gpu = n_gpu
LOGGER.info("device: {} n_gpu: {}, rank: {}, "
"16-bits training: {}".format(device, n_gpu, hvd.rank(),
opts.fp16))
if hvd.rank() != 0:
LOGGER.disabled = True
set_random_seed(opts.seed)
# train_examples = None
LOGGER.info(f"Loading the whole video dataset {opts.sub_txt_db}, "
f"{opts.vfeat_db}")
if opts.task != "didemo_video_only":
video_db = load_video_sub_dataset(opts.vfeat_db, opts.sub_txt_db,
opts.vfeat_interval, opts)
else:
txt_meta = load_json(join(opts.train_query_txt_db, "meta.json"))
video_db = load_video_only_dataset(opts.vfeat_db, txt_meta,
opts.vfeat_interval, opts)
# data loaders
# train
video_ids = get_video_ids(opts.train_query_txt_db)
train_q_txt_db = QueryTokLmdb(opts.train_query_txt_db, opts.max_txt_len)
train_dataloaders = build_downstream_dataloaders([opts.task],
video_db,
video_ids,
True,
opts,
shuffle=True,
q_txt_db=train_q_txt_db)
meta_loader = MetaLoader(train_dataloaders,
accum_steps=opts.gradient_accumulation_steps,
distributed=n_gpu > 1)
meta_loader = PrefetchLoader(meta_loader)
# val
video_ids = get_video_ids(opts.val_query_txt_db)
val_q_txt_db = QueryTokLmdb(opts.val_query_txt_db, -1)
val_dataloaders = build_downstream_dataloaders([opts.task],
video_db,
video_ids,
False,
opts,
q_txt_db=val_q_txt_db)
if opts.task != "didemo_video_only":
inf_dataset = VcmrFullEvalDataset
else:
inf_dataset = VcmrVideoOnlyFullEvalDataset
LOGGER.info(f"Loading Inference Dataset {opts.val_query_txt_db} (val)")
val_dset = inf_dataset(video_ids,
video_db,
val_q_txt_db,
distributed=opts.distributed_eval)
inf_loader_val = DataLoader(val_dset,
batch_size=opts.vcmr_eval_q_batch_size,
num_workers=opts.n_workers,
pin_memory=opts.pin_mem,
collate_fn=vcmr_full_eval_collate)
inf_loader_val = PrefetchLoader(inf_loader_val)
if opts.test_query_txt_db:
LOGGER.info(
f"Loading Inference Dataset {opts.test_query_txt_db} (test)")
video_ids = get_video_ids(opts.test_query_txt_db)
test_q_txt_db = QueryTokLmdb(opts.test_query_txt_db, -1)
test_dset = inf_dataset(video_ids,
video_db,
test_q_txt_db,
distributed=opts.distributed_eval)
inf_loader_test = DataLoader(test_dset,
batch_size=opts.vcmr_eval_q_batch_size,
num_workers=opts.n_workers,
pin_memory=opts.pin_mem,
collate_fn=vcmr_full_eval_collate)
inf_loader_test = PrefetchLoader(inf_loader_test)
# Prepare model
if opts.checkpoint:
checkpoint = torch.load(opts.checkpoint)
else:
checkpoint = {}
img_pos_embed_weight_key = "v_encoder.f_encoder.img_embeddings" +\
".position_embeddings.weight"
if img_pos_embed_weight_key in checkpoint:
max_frm_seq_len = len(checkpoint[img_pos_embed_weight_key])
else:
max_frm_seq_len = MAX_FRM_SEQ_LEN
model = HeroForVcmr.from_pretrained(
opts.model_config,
state_dict=checkpoint,
vfeat_dim=VFEAT_DIM,
max_frm_seq_len=max_frm_seq_len,
lw_neg_ctx=opts.lw_neg_ctx,
lw_neg_q=opts.lw_neg_q,
lw_st_ed=0,
ranking_loss_type=opts.ranking_loss_type,
use_hard_negative=False,
hard_pool_size=opts.hard_pool_size,
margin=opts.margin,
use_all_neg=opts.use_all_neg,
drop_svmr_prob=opts.drop_svmr_prob)
model.to(device)
# make sure every process has same model parameters in the beginning
broadcast_tensors([p.data for p in model.parameters()], 0)
set_dropout(model, opts.dropout)
# Prepare optimizer
optimizer = build_optimizer(model, opts)
task2scaler = {t: i for i, t in enumerate(train_dataloaders.keys())}
model, optimizer = amp.initialize(model,
optimizer,
num_losses=len(task2scaler),
enabled=opts.fp16,
opt_level='O2')
restorer = TrainingRestorer(opts, model, optimizer)
global_step = restorer.global_step
TB_LOGGER.global_step = global_step
if hvd.rank() == 0:
save_training_meta(opts)
TB_LOGGER.create(join(opts.output_dir, 'log'))
pbar = tqdm(total=opts.num_train_steps)
model_saver = ModelSaver(join(opts.output_dir, 'ckpt'))
if not exists(join(opts.output_dir, 'results')):
# store tvr predictions
os.makedirs(join(opts.output_dir, 'results'))
if opts.nms_thd != -1:
# store tvr-nms predictions
if not exists(join(opts.output_dir, 'results_nms')):
os.makedirs(join(opts.output_dir, 'results_nms'))
add_log_to_file(join(opts.output_dir, 'log', 'log.txt'))
else:
pbar = NoOp()
model_saver = NoOp()
restorer = NoOp()
if global_step > 0:
pbar.update(global_step)
LOGGER.info(f"***** Running training with {n_gpu} GPUs *****")
LOGGER.info(" Batch size = %d", opts.train_batch_size)
LOGGER.info(" Accumulate steps = %d", opts.gradient_accumulation_steps)
LOGGER.info(" Num steps = %d", opts.num_train_steps)
task2loss = {
task: RunningMeter(f'loss/{task}')
for task in train_dataloaders.keys()
}
for obj in (f'{opts.task}_st_ed', f'{opts.task}_neg_ctx',
f'{opts.task}_neg_q'):
task2loss[obj] = RunningMeter(f'loss/{obj}')
model.train()
n_examples = defaultdict(int)
start = time()
# quick hack for amp delay_unscale bug
optimizer.zero_grad()
if global_step == 0:
optimizer.step()
for step, (task, batch) in enumerate(meta_loader):
if len(opts.hard_negtiave_start_step) > 0:
for i, hn_step in enumerate(opts.hard_negtiave_start_step):
if global_step >= hn_step and hn_step != -1:
model.set_hard_negative(True, opts.hard_pool_size[i],
opts.hard_neg_weights[i])
if opts.train_span_start_step != -1 and\
global_step >= opts.train_span_start_step:
model.set_train_st_ed(opts.lw_st_ed)
n_examples[task] += opts.train_batch_size
loss = model(batch, task=task, compute_loss=True)
loss_st_ed, loss_neg_ctx, loss_neg_q = loss
loss = loss_st_ed + loss_neg_ctx + loss_neg_q
for n, ls, w in (('st_ed', loss_st_ed, opts.lw_st_ed),
('neg_ctx', loss_neg_ctx, opts.lw_neg_ctx),
('neg_q', loss_neg_q, opts.lw_neg_q)):
ls = ls.item()
if w:
ls /= w
task2loss[f'{task}_{n}'](ls)
loss = loss.mean()
task2loss[task](loss.item())
delay_unscale = (step + 1) % opts.gradient_accumulation_steps != 0
with amp.scale_loss(loss,
optimizer,
delay_unscale=delay_unscale,
loss_id=task2scaler[task]) as scaled_loss:
scaled_loss.backward()
if not delay_unscale:
# gather gradients from every processes
# do this before unscaling to make sure every process uses
# the same gradient scale
grads = [
p.grad.data for p in model.parameters()
if p.requires_grad and p.grad is not None
]
all_reduce_and_rescale_tensors(grads, float(1))
if (step + 1) % opts.gradient_accumulation_steps == 0:
global_step += 1
# learning rate scheduling
lr_this_step = get_lr_sched(global_step, opts)
for param_group in optimizer.param_groups:
param_group['lr'] = lr_this_step
TB_LOGGER.add_scalar('lr', lr_this_step, global_step)
# log loss
TB_LOGGER.log_scaler_dict({
temp_loss.name: temp_loss.val
for temp_loss in task2loss.values()
if temp_loss.val is not None
})
TB_LOGGER.step()
# update model params
if opts.grad_norm != -1:
grad_norm = clip_grad_norm_(amp.master_params(optimizer),
opts.grad_norm)
TB_LOGGER.add_scalar('grad_norm', grad_norm, global_step)
optimizer.step()
optimizer.zero_grad()
pbar.update(1)
if global_step % 100 == 0:
# monitor training throughput
LOGGER.info('-------------------------------------------')
LOGGER.info(f'Step {global_step}:')
for t in train_dataloaders.keys():
tot_ex = sum(all_gather_list(n_examples[t]))
ex_per_sec = int(tot_ex / (time() - start))
LOGGER.info(f'{t}: {tot_ex} examples trained at '
f'{ex_per_sec} ex/s')
TB_LOGGER.add_scalar(f'perf/{t}_ex_per_s', ex_per_sec,
global_step)
if global_step % opts.valid_steps == 0:
LOGGER.info('===========================================')
LOGGER.info(f"Step {global_step}: start running validation")
validate(model, val_dataloaders, opts)
if hvd.rank() == 0 or opts.distributed_eval:
log, results = validate_full_vcmr(model,
inf_loader_val,
'val',
opts,
model_opts=opts)
save_json(
results, f'{opts.output_dir}/results/'
f'val_results_{global_step}_rank{hvd.rank()}.json')
TB_LOGGER.log_scaler_dict(log)
if opts.test_query_txt_db:
log, results = validate_full_vcmr(model,
inf_loader_test,
'test',
opts,
model_opts=opts)
save_json(
results, f'{opts.output_dir}/results/'
f'test_results_{global_step}_rank{hvd.rank()}.json'
)
TB_LOGGER.log_scaler_dict(log)
LOGGER.info('===========================================')
model_saver.save(model, global_step)
# step restorer in the end to prevent missing validation checkpoint
restorer.step()
if global_step >= opts.num_train_steps:
break
LOGGER.info('===========================================')
if global_step % opts.valid_steps != 0:
if hvd.rank() == 0 or opts.distributed_eval:
log, results = validate_full_vcmr(model,
inf_loader_val,
'val',
opts,
model_opts=opts)
save_json(
results, f'{opts.output_dir}/results/'
f'val_results_{global_step}'
f'_rank{hvd.rank()}_final.json')
TB_LOGGER.log_scaler_dict(log)
if opts.test_query_txt_db:
log, results = validate_full_vcmr(model,
inf_loader_test,
'test',
opts,
model_opts=opts)
save_json(
results, f'{opts.output_dir}/results/'
f'test_results_{global_step}_rank{hvd.rank()}.json')
TB_LOGGER.log_scaler_dict(log)
model_saver.save(model, f'{global_step}_final')
def main(opts):
hvd.init()
n_gpu = hvd.size()
device = torch.device("cuda", hvd.local_rank())
torch.cuda.set_device(hvd.local_rank())
rank = hvd.rank()
opts.rank = rank
LOGGER.info("device: {} n_gpu: {}, rank: {}, "
"16-bits training: {}".format(
device, n_gpu, hvd.rank(), opts.fp16))
if opts.gradient_accumulation_steps < 1:
raise ValueError("Invalid gradient_accumulation_steps parameter: {}, "
"should be >= 1".format(
opts.gradient_accumulation_steps))
set_random_seed(opts.seed)
if rank == 0:
save_training_meta(opts)
TB_LOGGER.create(join(opts.output_dir, 'log'))
pbar = tqdm(total=opts.num_train_steps)
model_saver = ModelSaver(join(args.output_dir, 'ckpt'))
add_log_to_file(join(opts.output_dir, 'log', 'log.txt'))
else:
LOGGER.disabled = True
pbar = NoOp()
model_saver = NoOp()
all_dbs = [db for datasets in [opts.train_datasets, opts.val_datasets]
for dset in datasets for db in dset['db']]
tokenizer = json.load(open(f'{all_dbs[0]}/meta.json'))['bert']
assert all(tokenizer == json.load(open(f'{db}/meta.json'))['bert']
for db in all_dbs)
# build data loaders
train_dataloaders, all_img_dbs = create_dataloaders(
opts.train_datasets, True, opts)
val_dataloaders, _ = create_dataloaders(
opts.val_datasets, False, opts, all_img_dbs)
meta_loader = MetaLoader(train_dataloaders,
accum_steps=opts.gradient_accumulation_steps,
distributed=n_gpu > 1)
meta_loader = PrefetchLoader(meta_loader)
# Prepare model
if opts.checkpoint:
checkpoint = torch.load(opts.checkpoint)
else:
checkpoint = {}
model = UniterForPretraining.from_pretrained(
opts.model_config, checkpoint,
img_dim=IMG_DIM, img_label_dim=IMG_LABEL_DIM)
model.to(device)
model.train()
# make sure every process has same model parameters in the beginning
broadcast_tensors([p.data for p in model.parameters()], 0)
set_dropout(model, opts.dropout)
# Prepare optimizer
optimizer = build_optimizer(model, opts)
task2scaler = {t: i for i, t in enumerate(train_dataloaders.keys())}
model, optimizer = amp.initialize(model, optimizer,
num_losses=len(task2scaler),
enabled=opts.fp16, opt_level='O2')
global_step = 0
LOGGER.info(f"***** Running training with {n_gpu} GPUs *****")
LOGGER.info(" Batch size = %d", opts.train_batch_size)
LOGGER.info(" Accumulate steps = %d", opts.gradient_accumulation_steps)
LOGGER.info(" Num steps = %d", opts.num_train_steps)
# to compute training statistics
task2loss = {task: RunningMeter(f'loss/{task}')
for task in train_dataloaders.keys()}
# ITM w/ OT
if opts.itm_ot_lambda > 0:
for task in train_dataloaders.keys():
if task.startswith('itm'):
task2loss[f'{task}_xe'] = RunningMeter(f'loss/{task}_xe')
task2loss[f'{task}_ot'] = RunningMeter(f'loss/{task}_ot')
task2loss[f'{task}_ot_pos'] = RunningMeter(
f'loss/{task}_ot_pos')
task2loss[f'{task}_ot_neg'] = RunningMeter(
f'loss/{task}_ot_neg')
n_examples = defaultdict(int)
n_in_units = defaultdict(int)
n_loss_units = defaultdict(int)
grad_norm = 0
start = time()
# quick hack for amp delay_unscale bug
optimizer.zero_grad()
optimizer.step()
for step, (name, batch) in enumerate(meta_loader):
# forward pass
n_examples[name] += batch['input_ids'].size(0)
n_in_units[name] += (batch['attn_masks'] == 1).sum().item()
task = name.split('_')[0]
loss = model(batch, task=task, compute_loss=True)
if task.startswith('itm'):
# OT
itm_loss, ot_loss = loss
n_loss_units[name] += itm_loss.size(0)
itm_loss = itm_loss.mean()
if ot_loss is not None:
ot_pos, ot_neg = ot_loss
ot_loss = (ot_pos.sum() - ot_neg.sum()
) / (ot_pos.size(0) + ot_neg.size(0))
# NOTE: be ware of empty tensor
ot_pos = ot_pos.mean().item()
if not math.isnan(ot_pos):
task2loss[f'{name}_ot_pos'](ot_pos)
ot_neg = ot_neg.mean().item()
if not math.isnan(ot_neg):
task2loss[f'{name}_ot_neg'](ot_neg)
loss = itm_loss + opts.itm_ot_lambda * ot_loss
task2loss[f'{name}_xe'](itm_loss.item())
task2loss[f'{name}_ot'](ot_loss.item())
else:
loss = itm_loss
else:
n_loss_units[name] += loss.size(0)
loss = loss.mean() # loss is not normalized in model
# backward pass
delay_unscale = (step+1) % opts.gradient_accumulation_steps != 0
with amp.scale_loss(loss, optimizer, delay_unscale=delay_unscale,
loss_id=task2scaler[name]) as scaled_loss:
scaled_loss.backward()
if not delay_unscale:
# gather gradients from every processes
# do this before unscaling to make sure every process uses
# the same gradient scale
grads = [p.grad.data for p in model.parameters()
if p.requires_grad and p.grad is not None]
all_reduce_and_rescale_tensors(grads, float(1))
task2loss[name](loss.item())
# optimizer update and logging
if (step + 1) % opts.gradient_accumulation_steps == 0:
global_step += 1
# learning rate scheduling
lr_this_step = get_lr_sched(global_step, opts)
for param_group in optimizer.param_groups:
param_group['lr'] = lr_this_step
TB_LOGGER.add_scalar('lr', lr_this_step, global_step)
# log loss
# NOTE: not gathered across GPUs for efficiency
TB_LOGGER.log_scaler_dict({ll.name: ll.val
for ll in task2loss.values()
if ll.val is not None})
TB_LOGGER.step()
# update model params
if opts.grad_norm != -1:
grad_norm = clip_grad_norm_(amp.master_params(optimizer),
opts.grad_norm)
TB_LOGGER.add_scalar('grad_norm', grad_norm, global_step)
optimizer.step()
optimizer.zero_grad()
pbar.update(1)
if global_step % 100 == 0:
# monitor training throughput
LOGGER.info(f'==============Step {global_step}===============')
for t in train_dataloaders.keys():
assert all(tt == t for tt in all_gather_list(t))
tot_ex = sum(all_gather_list(n_examples[t]))
ex_per_sec = int(tot_ex / (time()-start))
tot_in = sum(all_gather_list(n_in_units[t]))
in_per_sec = int(tot_in / (time()-start))
tot_l = sum(all_gather_list(n_loss_units[t]))
l_per_sec = int(tot_l / (time()-start))
LOGGER.info(f'{t}: {tot_ex} examples trained at '
f'{ex_per_sec} ex/s')
TB_LOGGER.add_scalar(f'perf/{t}_ex_per_s', ex_per_sec,
global_step)
TB_LOGGER.add_scalar(f'perf/{t}_in_per_s', in_per_sec,
global_step)
TB_LOGGER.add_scalar(f'perf/{t}_loss_per_s', l_per_sec,
global_step)
LOGGER.info('===============================================')
if global_step % opts.valid_steps == 0:
LOGGER.info(f'Step {global_step}: start validation')
validate(model, val_dataloaders)
model_saver.save(model, global_step)
if global_step >= opts.num_train_steps:
break
if global_step % opts.valid_steps != 0:
LOGGER.info(f'Step {global_step}: start validation')
validate(model, val_dataloaders)
model_saver.save(model, global_step)
def main(opts):
hvd.init()
n_gpu = hvd.size()
device = torch.device("cuda", hvd.local_rank())
torch.cuda.set_device(hvd.local_rank())
rank = hvd.rank()
opts.rank = rank
LOGGER.info("device: {} n_gpu: {}, rank: {}, "
"16-bits training: {}".format(device, n_gpu, hvd.rank(),
opts.fp16))
if opts.gradient_accumulation_steps < 1:
raise ValueError("Invalid gradient_accumulation_steps parameter: {}, "
"should be >= 1".format(
opts.gradient_accumulation_steps))
set_random_seed(opts.seed)
if hvd.rank() == 0:
save_training_meta(opts)
TB_LOGGER.create(join(opts.output_dir, 'log'))
pbar = tqdm(total=opts.num_train_steps)
model_saver = ModelSaver(join(opts.output_dir, 'ckpt'))
os.makedirs(join(opts.output_dir, 'ckpt'), exist_ok=True)
add_log_to_file(join(opts.output_dir, 'log', 'log.txt'))
# store ITM predictions
os.makedirs(join(opts.output_dir, 'results_val'), exist_ok=True)
os.makedirs(join(opts.output_dir, 'results_test'), exist_ok=True)
os.makedirs(join(opts.output_dir, 'results_train'), exist_ok=True)
else:
LOGGER.disabled = True
pbar = NoOp()
model_saver = NoOp()
# train_examples = None
LOGGER.info(f"Loading Train Dataset {opts.train_txt_dbs}, "
f"{opts.train_img_dbs}")
# check multiple DBs
assert len(opts.train_txt_dbs) == len(opts.train_img_dbs), \
"train txt_db and img_db have different length"
# load DBs and image dirs
all_img_dbs = ImageLmdbGroup(opts.conf_th, opts.max_bb, opts.min_bb,
opts.num_bb, opts.compressed_db)
# train
LOGGER.info(f"Loading Train Dataset "
f"{opts.train_txt_dbs}, {opts.train_img_dbs}")
train_datasets = []
for txt_path, img_path in zip(opts.train_txt_dbs, opts.train_img_dbs):
if "itm_coco_zh" not in txt_path:
img_db = all_img_dbs[img_path]
txt_db = TxtTokLmdb(txt_path, opts.max_txt_len)
if opts.hard_neg_size > 0:
train_datasets.append(
ItmRankDatasetHardNeg(txt_db, img_db, opts.negative_size,
opts.hard_neg_size))
else:
train_datasets.append(
ItmRankDataset(txt_db, img_db, opts.negative_size))
else:
img_train_db = all_img_dbs[img_path[0]]
img_val_db = all_img_dbs[img_path[1]]
txt_db = TxtTokLmdb(txt_path, opts.max_txt_len)
if opts.hard_neg_size > 0:
train_datasets.append(
ItmRankDatasetHardNeg(txt_db, img_db, opts.negative_size,
opts.hard_neg_size))
else:
train_datasets.append(
ItmRankDataset_COCO_CN(txt_db, img_train_db, img_val_db,
opts.negative_size))
train_dataset = ConcatDataset(train_datasets)
# hard negative
# hn_datasets = []
# for txt_path, img_path in zip(opts.train_txt_dbs, opts.train_img_dbs):
# img_db = all_img_dbs[img_path]
# txt_db = TxtTokLmdb(txt_path, opts.max_txt_len)
# hn_datasets.append(ItmHardNegDataset(txt_db, img_db,
# opts.inf_minibatch_size))
# hn_dataset = ConcatDataset(hn_datasets)
# hn_dataloader = build_dataloader(hn_dataset, itm_hn_collate, False, opts)
# hard_neg_dir = f'{opts.output_dir}/results_train/'
# val
LOGGER.info(f"Loading Val Dataset {opts.val_txt_db}, {opts.val_img_db}")
val_img_db = all_img_dbs[opts.val_img_db[0]]
val_txt_db = TxtTokLmdb(opts.val_txt_db[0], -1)
val_dataset = ItmValDataset(val_txt_db, val_img_db,
opts.inf_minibatch_size)
val_dataloader = build_dataloader(val_dataset, itm_val_collate, False,
opts)
# eval
LOGGER.info(f"Loading val, test Dataset for full evaluation: "
f"{opts.val_txt_db}, {opts.val_img_db}"
f"{opts.test_txt_db}, {opts.test_img_db}")
eval_dataset_val = ItmEvalDataset(val_txt_db, val_img_db,
opts.inf_minibatch_size)
eval_loader_val = build_dataloader(eval_dataset_val, itm_eval_collate,
False, opts)
eval_loader_list = []
assert len(opts.test_img_db) == len(opts.test_txt_db)
for test_img_db_path, test_txt_db_path in zip(opts.test_img_db,
opts.test_txt_db):
if "itm_coco_zh" not in test_txt_db_path:
test_img_db = all_img_dbs[test_img_db_path]
test_txt_db = TxtTokLmdb(test_txt_db_path, -1)
eval_dataset_test = ItmEvalDataset(test_txt_db, test_img_db,
opts.inf_minibatch_size)
else:
test_img_train_db = all_img_dbs[test_img_db_path[0]]
test_img_val_db = all_img_dbs[test_img_db_path[1]]
test_txt_db = TxtTokLmdb(test_txt_db_path, -1)
eval_dataset_test = ItmEvalDataset_COCO_CN(test_txt_db,
test_img_train_db,
test_img_val_db,
opts.inf_minibatch_size)
eval_loader_test = build_dataloader(eval_dataset_test,
itm_eval_collate, False, opts)
eval_loader_list.append(eval_loader_test)
# Prepare model
if opts.checkpoint:
checkpoint = torch.load(opts.checkpoint)
else:
checkpoint = {}
#Rename the key if specified
if opts.rename_checkpoints:
rename_checkpoint(checkpoint)
model = VLXLMRForImageTextRetrieval.from_pretrained(
opts.model_config,
state_dict=checkpoint,
load_embedding_only=opts.load_embedding_only,
load_layer=opts.load_layer,
img_dim=IMG_DIM,
margin=opts.margin)
model.init_output() # pretrain ITM head is different from ranking head
model.to(device)
# make sure every process has same model parameters in the beginning
broadcast_tensors([p.data for p in model.parameters()], 0)
set_dropout(model, opts.dropout)
# Prepare optimizer
if opts.separate_lr:
optimizer = build_xlmr_optimizer(model, opts)
else:
optimizer = build_optimizer(model, opts)
model, optimizer = amp.initialize(model,
optimizer,
enabled=opts.fp16,
opt_level='O2')
#global_step = 0
LOGGER.info(f"***** Running training on {n_gpu} GPUs *****")
LOGGER.info(" Num examples = %d", len(train_dataset) * hvd.size())
LOGGER.info(" Batch size = %d", opts.train_batch_size)
LOGGER.info(" Accumulate steps = %d", opts.gradient_accumulation_steps)
LOGGER.info(" Num steps = %d", opts.num_train_steps)
running_loss = RunningMeter('loss')
model.train()
if opts.steps_per_hard_neg != -1:
compute_hard_neg(model, hn_dataloader, train_dataset,
opts.hard_neg_pool_size, hard_neg_dir)
#Initialize the TrainingRestorer
restorer = TrainingRestorer(opts, model, optimizer)
global_step = restorer.global_step
TB_LOGGER._global_step = global_step
if hvd.rank() != 0:
restorer = NoOp() #Added for Restoring the Checkpoints
if global_step > 0:
pbar.update(global_step)
n_examples = 0
start = time()
# quick hack for amp delay_unscale bug
optimizer.zero_grad()
optimizer.step()
while True:
train_dataloader = build_dataloader(train_dataset,
xlmr_itm_rank_collate, True, opts)
for step, batch in enumerate(train_dataloader):
#print(batch['input_ids'])
n_examples += batch['input_ids'].size(0)
loss = model(batch, compute_loss=True)
loss = loss.mean()
delay_unscale = (step + 1) % opts.gradient_accumulation_steps != 0
with amp.scale_loss(loss, optimizer,
delay_unscale=delay_unscale) as scaled_loss:
scaled_loss.backward()
if not delay_unscale:
# gather gradients from every processes
# do this before unscaling to make sure every process uses
# the same gradient scale
grads = [
p.grad.data for p in model.parameters()
if p.requires_grad and p.grad is not None
]
all_reduce_and_rescale_tensors(grads, float(1))
running_loss(loss.item())
# print("run the loss")
if (step + 1) % opts.gradient_accumulation_steps == 0:
global_step += 1
# learning rate scheduling
lr_this_step = get_lr_sched(global_step, opts)
if opts.separate_lr:
#added by Mingyang
xlmr_lr_this_step = get_xlmr_lr_sched(global_step, opts)
for i, param_group in enumerate(optimizer.param_groups):
if i < 2:
param_group['lr'] = xlmr_lr_this_step
else:
param_group['lr'] = lr_this_step
TB_LOGGER.add_scalar('xlmr_lr', xlmr_lr_this_step,
global_step)
else:
for param_group in optimizer.param_groups:
param_group['lr'] = lr_this_step
TB_LOGGER.add_scalar('lr', lr_this_step, global_step)
# log loss
losses = all_gather_list(running_loss)
running_loss = RunningMeter(
'loss',
sum(l.val for l in losses) / len(losses))
TB_LOGGER.add_scalar('loss', running_loss.val, global_step)
TB_LOGGER.step()
# update model params
if opts.grad_norm != -1:
grad_norm = clip_grad_norm_(amp.master_params(optimizer),
opts.grad_norm)
TB_LOGGER.add_scalar('grad_norm', grad_norm, global_step)
optimizer.step()
optimizer.zero_grad()
pbar.update(1)
if global_step % 100 == 0:
# monitor training throughput
LOGGER.info(f'============Step {global_step}=============')
tot_ex = sum(all_gather_list(n_examples))
ex_per_sec = int(tot_ex / (time() - start))
LOGGER.info(f'{tot_ex} examples trained at '
f'{ex_per_sec} ex/s')
TB_LOGGER.add_scalar('perf/ex_per_s', ex_per_sec,
global_step)
LOGGER.info(f'===========================================')
if global_step % opts.valid_steps == 0 and global_step > 0:
# if global_step > 7000:
if opts.full_val:
val_log = evaluate(model, eval_loader_val)
TB_LOGGER.log_scaler_dict(
{f"valid/{k}": v
for k, v in val_log.items()})
#Log the information
# LOGGER.info(
# f"========================= {split} ===========================\n"
# f"image retrieval R1: {eval_log['img_r1']*100:.2f},\n"
# f"image retrieval R5: {eval_log['img_r5']*100:.2f},\n"
# f"image retrieval R10: {eval_log['img_r10']*100:.2f}\n"
# f"text retrieval R1: {eval_log['txt_r1']*100:.2f},\n"
# f"text retrieval R5: {eval_log['txt_r5']*100:.2f},\n"
# f"text retrieval R10: {eval_log['txt_r10']*100:.2f}")
# LOGGER.info("=========================================================")
else:
val_log = validate(model, val_dataloader)
TB_LOGGER.log_scaler_dict(val_log)
model_saver.save(model, global_step)
restorer.step()
if (opts.steps_per_hard_neg != -1
and global_step % opts.steps_per_hard_neg == 0):
# sample hard negatives for training
compute_hard_neg(model, hn_dataloader, train_dataset,
opts.hard_neg_pool_size, hard_neg_dir)
# break to reconstruct loader
# for potential multi-worker issue (not sure)
break
if global_step >= opts.num_train_steps:
break
if global_step >= opts.num_train_steps:
break
# NOTE can no longer count epochs
pbar.close()
# final validation
# val_log = validate(model, val_dataloader)
# TB_LOGGER.log_scaler_dict(val_log)
model_saver.save(model, f'{global_step}_final')
for i, loader in enumerate(eval_loader_list):
split = "test_{}".format(i)
eval_log = evaluate(model, loader)
TB_LOGGER.log_scaler_dict(
{f"eval/{split}_{k}": v
for k, v in eval_log.items()})
if hvd.rank() != 0:
continue
LOGGER.info(
f"========================= {split} ===========================\n"
f"image retrieval R1: {eval_log['img_r1']*100:.2f},\n"
f"image retrieval R5: {eval_log['img_r5']*100:.2f},\n"
f"image retrieval R10: {eval_log['img_r10']*100:.2f}\n"
f"text retrieval R1: {eval_log['txt_r1']*100:.2f},\n"
f"text retrieval R5: {eval_log['txt_r5']*100:.2f},\n"
f"text retrieval R10: {eval_log['txt_r10']*100:.2f}")
LOGGER.info("=========================================================")
def main(opts):
hvd.init()
n_gpu = hvd.size()
device = torch.device("cuda", hvd.local_rank())
torch.cuda.set_device(hvd.local_rank())
rank = hvd.rank()
opts.rank = rank
LOGGER.info("device: {} n_gpu: {}, rank: {}, "
"16-bits training: {}".format(device, n_gpu, hvd.rank(),
opts.fp16))
set_random_seed(opts.seed)
if hvd.rank() == 0:
save_training_meta(opts)
TB_LOGGER.create(join(opts.output_dir, "log"))
pbar = tqdm(total=opts.num_train_steps)
model_saver = ModelSaver(join(opts.output_dir, "ckpt"))
add_log_to_file(join(opts.output_dir, "log", "log.txt"))
# store ITM predictions
os.makedirs(join(opts.output_dir, "results_val"))
os.makedirs(join(opts.output_dir, "results_test"))
os.makedirs(join(opts.output_dir, "results_train"))
else:
LOGGER.disabled = True
pbar = NoOp()
model_saver = NoOp()
# train_examples = None
LOGGER.info(f"Loading Train Dataset {opts.train_txt_dbs}, "
f"{opts.train_img_dbs}")
# check multiple DBs
assert len(opts.train_txt_dbs) == len(
opts.train_img_dbs), "train txt_db and img_db have different length"
# load DBs and image dirs
all_img_dbs = ImageLmdbGroup(opts.conf_th, opts.max_bb, opts.min_bb,
opts.num_bb, opts.compressed_db)
# train
LOGGER.info(f"Loading Train Dataset "
f"{opts.train_txt_dbs}, {opts.train_img_dbs}")
train_datasets_t = []
train_datasets_i = []
for txt_path, img_path in zip(opts.train_txt_dbs, opts.train_img_dbs):
img_db = all_img_dbs[img_path]
txt_db = TxtTokLmdb(txt_path, opts.max_txt_len)
train_datasets_t.append(
ItmRankDatasetHardNegFromText(txt_db, img_db, opts.negative_size))
train_datasets_i.append(
ItmRankDatasetHardNegFromImage(txt_db, img_db, opts.negative_size))
train_dataset_t = ConcatDataset(train_datasets_t)
train_dataset_i = ConcatDataset(train_datasets_i)
train_dataloader_t = build_dataloader(train_dataset_t, itm_rank_hn_collate,
True, opts)
train_dataloader_i = build_dataloader(train_dataset_i, itm_rank_hn_collate,
True, opts)
# val
LOGGER.info(f"Loading Val Dataset {opts.val_txt_db}, {opts.val_img_db}")
val_img_db = all_img_dbs[opts.val_img_db]
val_txt_db = TxtTokLmdb(opts.val_txt_db, -1)
val_dataset = ItmValDataset(val_txt_db, val_img_db,
opts.inf_minibatch_size)
val_dataloader = build_dataloader(val_dataset, itm_val_collate, False,
opts)
# eval
LOGGER.info(f"Loading val, test Dataset for full evaluation: "
f"{opts.val_txt_db}, {opts.val_img_db}"
f"{opts.test_txt_db}, {opts.test_img_db}")
eval_dataset_val = ItmEvalDataset(val_txt_db, val_img_db,
opts.inf_minibatch_size)
eval_loader_val = build_dataloader(eval_dataset_val, itm_eval_collate,
False, opts)
test_img_db = all_img_dbs[opts.test_img_db]
test_txt_db = TxtTokLmdb(opts.test_txt_db, -1)
eval_dataset_test = ItmEvalDataset(test_txt_db, test_img_db,
opts.inf_minibatch_size)
eval_loader_test = build_dataloader(eval_dataset_test, itm_eval_collate,
False, opts)
# Prepare model
if opts.checkpoint:
checkpoint = torch.load(opts.checkpoint)
else:
checkpoint = {}
model = UniterForImageTextRetrievalHardNeg.from_pretrained(
opts.model_config,
state_dict=checkpoint,
img_dim=IMG_DIM,
margin=opts.margin,
hard_size=opts.hard_neg_size,
)
model.init_output() # pretrain ITM head is different from ranking head
model.to(device)
# make sure every process has same model parameters in the beginning
broadcast_tensors([p.data for p in model.parameters()], 0)
set_dropout(model, opts.dropout)
# Prepare optimizer
optimizer = build_optimizer(model, opts)
model, optimizer = amp.initialize(model,
optimizer,
enabled=opts.fp16,
opt_level="O2")
LOGGER.info(f"***** Running training on {n_gpu} GPUs *****")
LOGGER.info(" Num examples = %d",
sum(all_gather_list(len(train_dataset_t))))
LOGGER.info(" Batch size = %d", opts.train_batch_size)
LOGGER.info(" Num steps = %d", opts.num_train_steps)
running_loss = RunningMeter("loss")
model.train()
global_step = 0
step = 0
n_examples = 0
n_hard_ex = 0
start = time()
train_iter_i = iter(train_dataloader_i)
# quick hack for amp delay_unscale bug
optimizer.zero_grad()
optimizer.step()
while True:
for batch in train_dataloader_t:
# hard text from image
try:
batch_i = next(train_iter_i)
except StopIteration:
train_iter_i = iter(train_dataloader_i)
batch_i = next(train_iter_i)
n_examples += batch_i["attn_masks"].size(0)
loss = model(batch_i, sample_from="i", compute_loss=True)
n_hard_ex += loss.numel()
loss = loss.mean() / opts.train_batch_size
with amp.scale_loss(loss, optimizer,
delay_unscale=True) as scaled_loss:
scaled_loss.backward()
# hard image from text
n_examples += batch["attn_masks"].size(0)
loss = model(batch, sample_from="t", compute_loss=True)
n_hard_ex += loss.numel()
# NOTE we use gradient accumulation to implemented train_batch_size
loss = loss.mean() / opts.train_batch_size
step += 1
delay_unscale = step % opts.train_batch_size != 0
with amp.scale_loss(loss, optimizer,
delay_unscale=delay_unscale) as scaled_loss:
scaled_loss.backward()
if not delay_unscale:
# gather gradients from every processes
# do this before unscaling to make sure every process uses
# the same gradient scale
grads = [
p.grad.data for p in model.parameters()
if p.requires_grad and p.grad is not None
]
all_reduce_and_rescale_tensors(grads, float(1))
running_loss(loss.item())
if step % opts.train_batch_size == 0:
global_step += 1
# learning rate scheduling
lr_this_step = get_lr_sched(global_step, opts)
for param_group in optimizer.param_groups:
param_group["lr"] = lr_this_step
TB_LOGGER.add_scalar("lr", lr_this_step, global_step)
# log loss
# NOTE: not gathered across GPUs for efficiency
TB_LOGGER.add_scalar("loss", running_loss.val, global_step)
TB_LOGGER.step()
# update model params
if opts.grad_norm != -1:
grad_norm = clip_grad_norm_(amp.master_params(optimizer),
opts.grad_norm)
TB_LOGGER.add_scalar("grad_norm", grad_norm, global_step)
optimizer.step()
optimizer.zero_grad()
pbar.update(1)
if global_step % 100 == 0:
# monitor training throughput
LOGGER.info(f"------------Step {global_step}-------------")
tot_ex = sum(all_gather_list(n_examples))
ex_per_sec = int(tot_ex / (time() - start))
tot_hn = sum(all_gather_list(n_hard_ex))
hn_per_sec = int(tot_hn / (time() - start))
LOGGER.info(f"{tot_ex} ({tot_hn}) examples (hard) "
f"trained at {ex_per_sec} ({hn_per_sec}) ex/s")
TB_LOGGER.add_scalar("perf/ex_per_s", ex_per_sec,
global_step)
TB_LOGGER.add_scalar("perf/hn_per_s", hn_per_sec,
global_step)
LOGGER.info(f"-------------------------------------------")
if global_step % opts.valid_steps == 0:
if opts.full_val:
LOGGER.info(
f"========================== Step {global_step} "
f"==========================")
val_log = evaluate(model, eval_loader_val)
TB_LOGGER.log_scaler_dict(
{f"valid/{k}": v
for k, v in val_log.items()})
LOGGER.info(f"image retrieval R1: "
f"{val_log['img_r1']*100:.2f},\n"
f"image retrieval R5: "
f"{val_log['img_r5']*100:.2f},\n"
f"image retrieval R10: "
f"{val_log['img_r10']*100:.2f}\n"
f"text retrieval R1: "
f"{val_log['txt_r1']*100:.2f},\n"
f"text retrieval R5: "
f"{val_log['txt_r5']*100:.2f},\n"
f"text retrieval R10: "
f"{val_log['txt_r10']*100:.2f}")
LOGGER.info("================================="
"=================================")
else:
val_log = validate(model, val_dataloader)
TB_LOGGER.log_scaler_dict(val_log)
model_saver.save(model, global_step)
if global_step >= opts.num_train_steps:
break
if global_step >= opts.num_train_steps:
break
pbar.close()
# final validation
val_log = validate(model, val_dataloader)
TB_LOGGER.log_scaler_dict(val_log)
model_saver.save(model, f"{global_step}_final")
# evaluation
for split, loader in [("val", eval_loader_val),
("test", eval_loader_test)]:
eval_log = evaluate(model, loader)
TB_LOGGER.log_scaler_dict(
{f"eval/{split}_{k}": v
for k, v in eval_log.items()})
if hvd.rank() != 0:
continue
LOGGER.info(
f"========================= {split} ===========================\n"
f"image retrieval R1: {eval_log['img_r1']*100:.2f},\n"
f"image retrieval R5: {eval_log['img_r5']*100:.2f},\n"
f"image retrieval R10: {eval_log['img_r10']*100:.2f}\n"
f"text retrieval R1: {eval_log['txt_r1']*100:.2f},\n"
f"text retrieval R5: {eval_log['txt_r5']*100:.2f},\n"
f"text retrieval R10: {eval_log['txt_r10']*100:.2f}")
LOGGER.info("=========================================================")
def _gradient_accumulation(self,
true_batchs,
normalization,
total_stats,
report_stats,
ratio=1.):
if self.grad_accum_count > 1:
self.model.zero_grad()
for batch in true_batchs:
# dec_state = None
src = make_features(batch, 'src')
_, src_lengths = batch.src
tgt = make_features(batch, 'tgt')
# reconstructor input
stgt = make_features(batch, 'stgt')
stgt = stgt.transpose(0, 1)
stgt = stgt.transpose(1, 2)
# if choose child randomly, make sequence different from the deep first traversal method
choice = random.randint(0, stgt.size(0) - 1)
stgt = stgt[choice][:-1]
structure = make_features(batch, 'structure')
structure = structure.transpose(0, 1)
structure = structure.transpose(1, 2)
# 2. F-prop all but generator.
if self.grad_accum_count == 1:
self.model.zero_grad()
outputs, attns, s_outputs, s_attns = \
self.model(src, tgt, stgt, structure, src_lengths)
# 3. Compute loss in shards for memory efficiency.
batch_stats = self.train_loss.sharded_compute_loss(
batch, (outputs, s_outputs),
stgt,
self.shard_size,
normalization,
ratio=ratio)
total_stats.update(batch_stats)
report_stats.update(batch_stats)
# 4. Update the parameters and statistics.
if self.grad_accum_count == 1:
# Multi GPU gradient gather
if self.n_gpu > 1:
grads = [
p.grad.data for p in self.model.parameters()
if p.requires_grad and p.grad is not None
]
all_reduce_and_rescale_tensors(grads, float(1))
self.optim.step()
# If truncated, don't backprop fully.
# TO CHECK
# if dec_state is not None:
# dec_state.detach()
if self.model.decoder.state is not None:
self.model.decoder.detach_state()
# in case of multi step gradient accumulation,
# update only after accum batches
if self.grad_accum_count > 1:
if self.n_gpu > 1:
grads = [
p.grad.data for p in self.model.parameters()
if p.requires_grad and p.grad is not None
]
all_reduce_and_rescale_tensors(grads, float(1))
self.optim.step()
def main(opts):
hvd.init()
n_gpu = hvd.size()
device = torch.device("cuda", hvd.local_rank())
torch.cuda.set_device(hvd.local_rank())
rank = hvd.rank()
opts.rank = rank
LOGGER.info("device: {} n_gpu: {}, rank: {}, "
"16-bits training: {}".format(
device, n_gpu, hvd.rank(), opts.fp16))
if opts.gradient_accumulation_steps < 1:
raise ValueError("Invalid gradient_accumulation_steps parameter: {}, "
"should be >= 1".format(
opts.gradient_accumulation_steps))
set_random_seed(opts.seed)
# train_examples = None
LOGGER.info(f"Loading Train Dataset {opts.train_txt_db}, "
f"{opts.train_img_db}")
if 'paired' in opts.model:
DatasetCls = Nlvr2PairedDataset
EvalDatasetCls = Nlvr2PairedEvalDataset
collate_fn = nlvr2_paired_collate
eval_collate_fn = nlvr2_paired_eval_collate
if opts.model == 'paired':
ModelCls = UniterForNlvr2Paired
elif opts.model == 'paired-attn':
ModelCls = UniterForNlvr2PairedAttn
else:
raise ValueError('unrecognized model type')
elif opts.model == 'triplet':
DatasetCls = Nlvr2TripletDataset
EvalDatasetCls = Nlvr2TripletEvalDataset
ModelCls = UniterForNlvr2Triplet
collate_fn = nlvr2_triplet_collate
eval_collate_fn = nlvr2_triplet_eval_collate
else:
raise ValueError('unrecognized model type')
# data loaders
train_dataloader = create_dataloader(opts.train_img_db, opts.train_txt_db,
opts.train_batch_size, True,
DatasetCls, collate_fn, opts)
val_dataloader = create_dataloader(opts.val_img_db, opts.val_txt_db,
opts.val_batch_size, False,
EvalDatasetCls, eval_collate_fn, opts)
test_dataloader = create_dataloader(opts.test_img_db, opts.test_txt_db,
opts.val_batch_size, False,
EvalDatasetCls, eval_collate_fn, opts)
# Prepare model
if opts.checkpoint:
ckpt = torch.load(opts.checkpoint)
checkpoint = {k.replace('bert', 'uniter'): v for k, v in ckpt.items()}
else:
checkpoint = {}
model = ModelCls.from_pretrained(opts.model_config, state_dict=checkpoint,
img_dim=IMG_DIM)
model.init_type_embedding()
model.to(device)
# make sure every process has same model parameters in the beginning
broadcast_tensors([p.data for p in model.parameters()], 0)
set_dropout(model, opts.dropout)
# Prepare optimizer
optimizer = build_optimizer(model, opts)
model, optimizer = amp.initialize(model, optimizer,
enabled=opts.fp16, opt_level='O2')
global_step = 0
if rank == 0:
save_training_meta(opts)
TB_LOGGER.create(join(opts.output_dir, 'log'))
pbar = tqdm(total=opts.num_train_steps)
model_saver = ModelSaver(join(opts.output_dir, 'ckpt'))
os.makedirs(join(opts.output_dir, 'results')) # store val predictions
add_log_to_file(join(opts.output_dir, 'log', 'log.txt'))
else:
LOGGER.disabled = True
pbar = NoOp()
model_saver = NoOp()
LOGGER.info(f"***** Running training with {n_gpu} GPUs *****")
LOGGER.info(" Num examples = %d", len(train_dataloader.dataset))
LOGGER.info(" Batch size = %d", opts.train_batch_size)
LOGGER.info(" Accumulate steps = %d", opts.gradient_accumulation_steps)
LOGGER.info(" Num steps = %d", opts.num_train_steps)
running_loss = RunningMeter('loss')
model.train()
n_examples = 0
n_epoch = 0
start = time()
# quick hack for amp delay_unscale bug
optimizer.zero_grad()
optimizer.step()
while True:
for step, batch in enumerate(train_dataloader):
targets = batch['targets']
n_examples += targets.size(0)
# ============================ Code for adversarial training =============
if opts.adv_training:
# initialize delta
txt_embeds_init = model.uniter.embeddings.word_embeddings(
batch['input_ids'])
img_embeds_init = batch['img_feat']
# for simplicity, we initialize the delta as zero vectors, which performs
# very simliar as initializing randomly using norm or uniform distributions
txt_delta = torch.zeros_like(txt_embeds_init)
img_delta = torch.zeros_like(img_embeds_init)
# calculate the prob. scores for clean samples
gt_answer_scores = model(batch, compute_loss=False)
gt_answer_prob = F.softmax(gt_answer_scores, dim=1)
gt_answer_logprob = F.log_softmax(gt_answer_scores, dim=1)
# the main loop
for astep in range(opts.adv_steps):
# (0) forward
if opts.adv_modality == ["text"]:
txt_delta.requires_grad_()
img_delta = torch.zeros_like(img_embeds_init)
elif opts.adv_modality == ["image"]:
img_delta.requires_grad_()
txt_delta = torch.zeros_like(txt_embeds_init)
else:
txt_delta.requires_grad_()
img_delta.requires_grad_()
if "alter" not in opts.adv_modality:
answer_scores = model(batch, adv_training = True,
adv_modality = opts.adv_modality, adv_delta_txt = txt_delta,
adv_delta_img = img_delta, compute_loss=False)
# BCE loss
bce_loss = F.cross_entropy(answer_scores,
batch['targets'], reduction='none')
bce_loss = bce_loss.mean()
# KL loss
answer_prob = F.softmax(answer_scores, dim=1)
answer_logprob = F.log_softmax(answer_scores, dim=1)
kl_loss = F.kl_div(answer_logprob,gt_answer_prob,reduction='none') + \
F.kl_div(gt_answer_logprob,answer_prob,reduction='none')
kl_loss = kl_loss.sum(1).mean()
# (1) backward
loss = (bce_loss + opts.adv_kl_weight * kl_loss) / opts.adv_steps
else:
answer_scores_1 = model(batch, adv_training = True,
adv_modality = ["text"], adv_delta_txt = txt_delta,
adv_delta_img = None, compute_loss=False)
bce_loss_1 = F.cross_entropy(answer_scores_1,
batch['targets'], reduction='none')
bce_loss_1 = bce_loss_1.mean()
answer_scores_2 = model(batch, adv_training = True,
adv_modality = ["image"], adv_delta_txt = None,
adv_delta_img = img_delta, compute_loss=False)
bce_loss_2 = F.cross_entropy(answer_scores_2,
batch['targets'], reduction='none')
bce_loss_2 = bce_loss_2.mean()
# KL loss
answer_prob_1 = F.softmax(answer_scores_1, dim=1)
answer_logprob_1 = F.log_softmax(answer_scores_1, dim=1)
answer_prob_2 = F.softmax(answer_scores_2, dim=1)
answer_logprob_2 = F.log_softmax(answer_scores_2, dim=1)
kl_loss_1 = F.kl_div(answer_logprob_1,gt_answer_prob,reduction='none') + \
F.kl_div(gt_answer_logprob,answer_prob_1,reduction='none')
kl_loss_1 = kl_loss_1.sum(1).mean()
kl_loss_2 = F.kl_div(answer_logprob_2,gt_answer_prob,reduction='none') + \
F.kl_div(gt_answer_logprob,answer_prob_2,reduction='none')
kl_loss_2 = kl_loss_2.sum(1).mean()
# (1) backward
loss = (bce_loss_1 + bce_loss_2 + opts.adv_kl_weight * (kl_loss_1+kl_loss_2)) / (opts.adv_steps*2)
delay_unscale = ((step+1) % opts.gradient_accumulation_steps != 0) or ((astep+1) % opts.adv_steps != 0)
with amp.scale_loss(loss, optimizer, delay_unscale=delay_unscale
) as scaled_loss:
scaled_loss.backward(retain_graph=True)
if not delay_unscale:
# gather gradients from every processes
# do this before unscaling to make sure every process uses
# the same gradient scale
grads = [p.grad.data for p in model.parameters()
if p.requires_grad and p.grad is not None]
all_reduce_and_rescale_tensors(grads, float(1))
running_loss(loss.item())
if astep == opts.adv_steps - 1:
# further updates on delta
break
# (2) get gradient on delta
# fix fp16 problem
amp_scale = scaled_loss.item() // loss.item()
if "text" in opts.adv_modality:
txt_delta_grad = txt_delta.grad.clone().detach().float() / amp_scale
if "image" in opts.adv_modality:
img_delta_grad = img_delta.grad.clone().detach().float() / amp_scale
# (3) update and clip for txt delta
if "text" in opts.adv_modality:
if opts.norm_type == "l2":
denorm = torch.norm(txt_delta_grad.view(txt_delta_grad.size(0), -1), dim=1).view(-1, 1, 1)
denorm = torch.clamp(denorm, min=1e-8)
txt_delta_step = (opts.adv_lr_txt * txt_delta_grad / denorm).to(txt_delta)
txt_delta = (txt_delta + txt_delta_step).detach()
if opts.adv_max_norm > 0:
delta_norm = torch.norm(txt_delta.view(txt_delta.size(0), -1), p=2, dim=1).detach()
exceed_mask = (delta_norm > opts.adv_max_norm).to(txt_embeds_init)
reweights = (opts.adv_max_norm / delta_norm * exceed_mask + (1-exceed_mask)).view(-1, 1, 1)
txt_delta = (txt_delta * reweights).detach()
elif opts.norm_type == "linf":
denorm = torch.norm(txt_delta_grad.view(txt_delta_grad.size(0), -1), dim=1, p=float("inf")).view(-1, 1, 1)
denorm = torch.clamp(denorm, min=1e-8)
txt_delta_step = (opts.adv_lr_txt * txt_delta_grad / denorm).to(txt_delta)
txt_delta = (txt_delta + txt_delta_step).detach()
if opts.adv_max_norm > 0:
txt_delta = torch.clamp(txt_delta, -opts.adv_max_norm, opts.adv_max_norm).detach()
# (4) update and clip for image delta
if "image" in opts.adv_modality:
if opts.norm_type == "l2":
denorm = torch.norm(img_delta_grad.view(img_delta_grad.size(0), -1), dim=1).view(-1, 1, 1)
denorm = torch.clamp(denorm, min=1e-8)
img_delta_step = (opts.adv_lr_img * img_delta_grad / denorm).to(img_delta)
img_delta = (img_delta + img_delta_step).detach()
if opts.adv_max_norm > 0:
delta_norm = torch.norm(img_delta.view(img_delta.size(0), -1), p=2, dim=1).detach()
exceed_mask = (delta_norm > opts.adv_max_norm).to(img_embeds_init)
reweights = (opts.adv_max_norm / delta_norm * exceed_mask + (1-exceed_mask)).view(-1, 1, 1)
img_delta = (img_delta * reweights).detach()
elif opts.norm_type == "linf":
denorm = torch.norm(img_delta_grad.view(img_delta_grad.size(0), -1), dim=1, p=float("inf")).view(-1, 1, 1)
denorm = torch.clamp(denorm, min=1e-8)
img_delta_step = (opts.adv_lr_img * img_delta_grad / denorm).to(img_delta)
img_delta = (img_delta + img_delta_step).detach()
if opts.adv_max_norm > 0:
img_delta = torch.clamp(img_delta, -opts.adv_max_norm, opts.adv_max_norm).detach()
else:
loss = model(batch, compute_loss=True)
loss = loss.mean()
delay_unscale = (step+1) % opts.gradient_accumulation_steps != 0
with amp.scale_loss(loss, optimizer, delay_unscale=delay_unscale
) as scaled_loss:
scaled_loss.backward()
if not delay_unscale:
# gather gradients from every processes
# do this before unscaling to make sure every process uses
# the same gradient scale
grads = [p.grad.data for p in model.parameters()
if p.requires_grad and p.grad is not None]
all_reduce_and_rescale_tensors(grads, float(1))
running_loss(loss.item())
# ============================ End ==========================
if (step + 1) % opts.gradient_accumulation_steps == 0:
global_step += 1
# learning rate scheduling
lr_this_step = get_lr_sched(global_step, opts)
for param_group in optimizer.param_groups:
param_group['lr'] = lr_this_step
TB_LOGGER.add_scalar('lr', lr_this_step, global_step)
# log loss
# NOTE: not gathered across GPUs for efficiency
TB_LOGGER.add_scalar('loss', running_loss.val, global_step)
TB_LOGGER.step()
# update model params
if opts.grad_norm != -1:
grad_norm = clip_grad_norm_(amp.master_params(optimizer),
opts.grad_norm)
TB_LOGGER.add_scalar('grad_norm', grad_norm, global_step)
optimizer.step()
optimizer.zero_grad()
pbar.update(1)
if global_step % 100 == 0:
# monitor training throughput
tot_ex = sum(all_gather_list(n_examples))
ex_per_sec = int(tot_ex / (time()-start))
LOGGER.info(f'Step {global_step}: '
f'{tot_ex} examples trained at '
f'{ex_per_sec} ex/s')
TB_LOGGER.add_scalar('perf/ex_per_s',
ex_per_sec, global_step)
if global_step % opts.valid_steps == 0:
for split, loader in [('val', val_dataloader),
('test', test_dataloader)]:
LOGGER.info(f"Step {global_step}: start running "
f"validation on {split} split...")
log, results = validate(model, loader, split)
with open(f'{opts.output_dir}/results/'
f'{split}_results_{global_step}_'
f'rank{rank}.csv', 'w') as f:
for id_, ans in results:
f.write(f'{id_},{ans}\n')
TB_LOGGER.log_scaler_dict(log)
model_saver.save(model, global_step)
if global_step >= opts.num_train_steps:
break
if global_step >= opts.num_train_steps:
break
n_epoch += 1
LOGGER.info(f"Step {global_step}: finished {n_epoch} epochs")
if opts.num_train_steps % opts.valid_steps != 0:
for split, loader in [('val', val_dataloader),
('test', test_dataloader)]:
LOGGER.info(f"Step {global_step}: start running "
f"validation on {split} split...")
log, results = validate(model, loader, split)
with open(f'{opts.output_dir}/results/'
f'{split}_results_{global_step}_'
f'rank{rank}.csv', 'w') as f:
for id_, ans in results:
f.write(f'{id_},{ans}\n')
TB_LOGGER.log_scaler_dict(log)
model_saver.save(model, global_step)
def _gradient_accumulation(self, true_batchs, normalization, total_stats,
report_stats, ratio=0.15, ratio2=0.05):
if self.grad_accum_count > 1:
self.model.zero_grad()
for batch in true_batchs:
target_size = batch.tgt.size(0)
# dec_state = None
src = make_features(batch, 'src')
_, src_lengths = batch.src
tgt = make_features(batch, 'tgt')
# reconstructor input
stgt = make_features(batch, 'stgt')
stgt = stgt.transpose(0, 1)
stgt = stgt.transpose(1, 2)
# randomly traversal sequence
choice = random.randint(0, stgt.size(0) - 1)
stgt = stgt[choice][:-1]
structure = make_features(batch, 'structure')
structure = structure.transpose(0, 1)
structure = structure.transpose(1, 2)
# bad code
mask = make_features(batch, 'mask')
mask = mask - 2
mask[mask <= 0] = 0
mask = mask.byte()
relation = make_features(batch, 'relation')
relation = relation.transpose(0, 1)
relation = relation[relation != 1]
# 2. F-prop all but generator.
if self.grad_accum_count == 1:
self.model.zero_grad()
outputs, attns, s_outputs, s_attns, p, rels = \
self.model(src, tgt, stgt, structure, mask, src_lengths)
# 3. Compute loss in shards for memory efficiency.
batch_stats = self.train_loss.sharded_compute_loss(
batch, (outputs, s_outputs), stgt, self.shard_size, normalization, ratio1=1-ratio,
ratio2=ratio)
if relation.size(0)>0:
relation_loss = self.train_relation_loss(rels, relation)
loss = (-p + relation_loss) / relation.size(0)
loss = loss * ratio2
loss.backward()
total_stats.update(batch_stats)
report_stats.update(batch_stats)
# 4. Update the parameters and statistics.
if self.grad_accum_count == 1:
# Multi GPU gradient gather
if self.n_gpu > 1:
grads = [p.grad.data for p in self.model.parameters()
if p.requires_grad
and p.grad is not None]
all_reduce_and_rescale_tensors(
grads, float(1))
self.optim.step()
# If truncated, don't backprop fully.
# TO CHECK
# if dec_state is not None:
# dec_state.detach()
if self.model.decoder.state is not None:
self.model.decoder.detach_state()
# in case of multi step gradient accumulation,
# update only after accum batches
if self.grad_accum_count > 1:
if self.n_gpu > 1:
grads = [p.grad.data for p in self.model.parameters()
if p.requires_grad
and p.grad is not None]
all_reduce_and_rescale_tensors(
grads, float(1))
self.optim.step()
def main(opts, checkpoint_dir=None, tuning=False):
from utils.logger import LOGGER, TB_LOGGER, RunningMeter, add_log_to_file
with logger.catch(reraise=True):
logger.info(f"{opts}")
if isinstance(opts, dict):
opts = edict(opts)
hvd.init()
n_gpu = hvd.size()
device = torch.device("cuda", hvd.local_rank())
torch.cuda.set_device(hvd.local_rank())
rank = hvd.rank()
opts.rank = rank
LOGGER.info("device: {} n_gpu: {}, rank: {}, "
"16-bits training: {}".format(device, n_gpu, hvd.rank(),
opts.fp16))
if opts.gradient_accumulation_steps < 1:
raise ValueError(
"Invalid gradient_accumulation_steps parameter: {}, "
"should be >= 1".format(opts.gradient_accumulation_steps))
set_random_seed(opts.seed)
"""
# load DBs and image dirs
"""
all_img_dbs = ImageLmdbGroup(opts.conf_th, opts.max_bb, opts.min_bb,
opts.num_bb, opts.compressed_db)
# train
LOGGER.info(f"Loading Train Dataset "
f"{opts.train_txt_dbs}, {opts.train_img_dbs}")
train_datasets = []
for txt_path, img_path in zip(opts.train_txt_dbs, opts.train_img_dbs):
img_db = all_img_dbs[img_path]
txt_db = TxtTokLmdb(txt_path, opts.max_txt_len)
train_datasets.append(MemeDataset(1, txt_db, img_db))
train_dataset = ConcatDatasetWithLens(train_datasets)
train_dataloader = build_dataloader(train_dataset, meme_collate, True,
opts)
# val
LOGGER.info(
f"Loading Train Dataset {opts.val_txt_db}, {opts.val_img_db}")
val_img_db = all_img_dbs[opts.val_img_db]
val_txt_db = TxtTokLmdb(opts.val_txt_db, -1)
val_dataset = MemeEvalDataset(1, val_txt_db, val_img_db)
val_dataloader = build_dataloader(val_dataset,
meme_eval_itm_ot_collate, False,
opts)
# test_img_db = val_img_db
# test_txt_db = TxtTokLmdb(opts.test_txt_db, -1)
# test_dataset = MemeEvalDataset(1, test_txt_db, test_img_db)
# test_dataloader = build_dataloader(test_dataset, meme_eval_collate,
# False, opts)
"""
# Prepare model
"""
if opts.checkpoint:
checkpoint = torch.load(opts.checkpoint)
else:
checkpoint = {}
all_dbs = opts.train_txt_dbs + [opts.val_txt_db]
model = UniterForITM.from_pretrained(opts.model_config,
checkpoint,
img_dim=IMG_DIM,
num_answer=1)
model.to(device)
# make sure every process has same model parameters in the beginning
broadcast_tensors([p.data for p in model.parameters()], 0)
set_dropout(model, opts.dropout)
"""
# Prepare optimizer
"""
optimizer = build_optimizer(model, opts)
model, optimizer = amp.initialize(model,
optimizer,
enabled=opts.fp16,
opt_level='O2')
global_step = 0
if rank == 0:
save_training_meta(opts)
TB_LOGGER.create(join(opts.output_dir, 'log'))
pbar = tqdm(total=opts.num_train_steps)
model_saver = ModelSaver(join(opts.output_dir, 'ckpt'))
# json.dump(ans2label,
# open(join(opts.output_dir, 'ckpt', 'ans2label.json'), 'w'))
os.makedirs(join(opts.output_dir, 'results'),
exist_ok=tuning) # store VQA predictions
add_log_to_file(join(opts.output_dir, 'log', 'log.txt'))
else:
LOGGER.disabled = True
pbar = NoOp()
model_saver = NoOp()
LOGGER.info(f"***** Running training with {n_gpu} GPUs *****")
LOGGER.info(" Num examples = %d", len(train_dataset) * hvd.size())
LOGGER.info(" Batch size = %d", opts.train_batch_size)
LOGGER.info(" Accumulate steps = %d",
opts.gradient_accumulation_steps)
LOGGER.info(" Num steps = %d", opts.num_train_steps)
running_loss = RunningMeter('loss')
model.train()
n_examples = 0
n_epoch = 0
if checkpoint_dir is not None and tuning:
checkpoint = os.path.join(checkpoint_dir, "checkpoint")
(model_state, optimizer_state, n_epoch,
n_examples) = torch.load(checkpoint)
model.load_state_dict(model_state)
optimizer.load_state_dict(optimizer_state)
LOGGER.info(
f"***** Resume from ray tune checkpoint : {checkpoint_dir} *****"
)
LOGGER.info(" n_examples = %d", n_examples)
LOGGER.info(" n_epoch = %d", n_epoch)
# shutil.rmtree(checkpoint_dir)
start = time()
# quick hack for amp delay_unscale bug
optimizer.zero_grad()
optimizer.step()
while True:
for step, batch in enumerate(train_dataloader):
if global_step > 2000:
logger.error('Force stop at global step 2000')
sys.exit(0)
n_examples += batch['input_ids'].size(0)
if opts.adv_training:
# NOTE: reverse label like what we do in UniterForITM
targets = batch['targets']
targets = (targets > 0.5).long()
targets = torch.abs(targets - 1)
batch['targets'] = targets
# initialize delta
txt_embeds_init = model.uniter.embeddings.word_embeddings(
batch['input_ids'])
img_embeds_init = batch['img_feat']
# for simplicity, we initialize the delta as zero vectors, which performs
# very simliar as initializing randomly using norm or uniform distributions
txt_delta = torch.zeros_like(txt_embeds_init)
img_delta = torch.zeros_like(img_embeds_init)
# calculate the prob. scores for clean samples
gt_answer_scores = model(batch, compute_loss=False)
gt_answer_prob = F.softmax(gt_answer_scores, dim=1)
gt_answer_logprob = F.log_softmax(gt_answer_scores, dim=1)
# the main loop
for astep in range(opts.adv_steps):
# (0) forward
if opts.adv_modality == ["text"]:
txt_delta.requires_grad_()
img_delta = torch.zeros_like(img_embeds_init)
elif opts.adv_modality == ["image"]:
img_delta.requires_grad_()
txt_delta = torch.zeros_like(txt_embeds_init)
else:
txt_delta.requires_grad_()
img_delta.requires_grad_()
if "alter" not in opts.adv_modality:
answer_scores = model(
batch,
adv_training=True,
adv_modality=opts.adv_modality,
adv_delta_txt=txt_delta,
adv_delta_img=img_delta,
compute_loss=False)
# CE loss
ce_loss = F.cross_entropy(
answer_scores,
batch['targets'].squeeze(-1),
reduction='mean')
# KL loss
answer_prob = F.softmax(answer_scores, dim=1)
answer_logprob = F.log_softmax(answer_scores,
dim=1)
kl_loss = F.kl_div(
answer_logprob, gt_answer_prob, reduction='none') + \
F.kl_div(
gt_answer_logprob, answer_prob,
reduction='none')
kl_loss = kl_loss.mean()
# (1) backward
loss = (ce_loss + opts.adv_kl_weight *
kl_loss) / opts.adv_steps
else:
answer_scores_1 = model(batch,
adv_training=True,
adv_modality=["text"],
adv_delta_txt=txt_delta,
adv_delta_img=None,
compute_loss=False)
# CE loss
ce_loss_1 = F.cross_entropy(
answer_scores,
batch['targets'].squeeze(-1),
reduction='mean')
answer_scores_2 = model(batch,
adv_training=True,
adv_modality=["image"],
adv_delta_txt=None,
adv_delta_img=img_delta,
compute_loss=False)
# CE loss
ce_loss_2 = F.cross_entropy(
answer_scores,
batch['targets'].squeeze(-1),
reduction='mean')
# KL loss
answer_prob_1 = F.softmax(answer_scores_1, dim=1)
answer_logprob_1 = F.log_softmax(answer_scores_1,
dim=1)
answer_prob_2 = F.softmax(answer_scores_2, dim=1)
answer_logprob_2 = F.log_softmax(answer_scores_2,
dim=1)
kl_loss_1 = F.kl_div(
answer_logprob_1, gt_answer_prob, reduction='none') + \
F.kl_div(
gt_answer_logprob, answer_prob_1,
reduction='none')
kl_loss_1 = kl_loss_1.mean()
kl_loss_2 = F.kl_div(
answer_logprob_2, gt_answer_prob, reduction='none') + \
F.kl_div(
gt_answer_logprob, answer_prob_2,
reduction='none')
kl_loss_2 = kl_loss_2.mean()
# (1) backward
loss = (
ce_loss_1 + ce_loss_2 + opts.adv_kl_weight *
(kl_loss_1 + kl_loss_2)) / (opts.adv_steps * 2)
delay_unscale = (
(step + 1) % opts.gradient_accumulation_steps !=
0) or ((astep + 1) % opts.adv_steps != 0)
with amp.scale_loss(
loss, optimizer,
delay_unscale=delay_unscale) as scaled_loss:
scaled_loss.backward(retain_graph=True)
if not delay_unscale:
# gather gradients from every processes
# do this before unscaling
# to make sure every process uses
# the same gradient scale
grads = [
p.grad.data for p in model.parameters()
if p.requires_grad and p.grad is not None
]
all_reduce_and_rescale_tensors(grads, float(1))
running_loss(loss.item())
if astep == opts.adv_steps - 1:
# further updates on delta
break
# (2) get gradient on delta
# fix fp16 problem
amp_scale = scaled_loss.item() // loss.item()
if "text" in opts.adv_modality:
txt_delta_grad = txt_delta.grad.clone().detach()
txt_delta_grad = txt_delta_grad.float() / amp_scale
if "image" in opts.adv_modality:
img_delta_grad = img_delta.grad.clone().detach()
img_delta_grad = img_delta_grad.float() / amp_scale
# (3) update and clip for txt delta
if "text" in opts.adv_modality:
if opts.norm_type == "l2":
denorm = torch.norm(txt_delta_grad.view(
txt_delta_grad.size(0), -1),
dim=1).view(-1, 1, 1)
denorm = torch.clamp(denorm, min=1e-8)
txt_delta_step = (opts.adv_lr_txt *
txt_delta_grad /
denorm).to(txt_delta)
txt_delta = (txt_delta +
txt_delta_step).detach()
if opts.adv_max_norm > 0:
delta_norm = torch.norm(txt_delta.view(
txt_delta.size(0), -1),
p=2,
dim=1).detach()
exceed_mask = (
delta_norm >
opts.adv_max_norm).to(txt_embeds_init)
reweights = (opts.adv_max_norm /
delta_norm * exceed_mask +
(1 - exceed_mask)).view(
-1, 1, 1)
txt_delta = (txt_delta *
reweights).detach()
elif opts.norm_type == "linf":
denorm = torch.norm(txt_delta_grad.view(
txt_delta_grad.size(0), -1),
dim=1,
p=float("inf")).view(
-1, 1, 1)
denorm = torch.clamp(denorm, min=1e-8)
txt_delta_step = (opts.adv_lr_txt *
txt_delta_grad /
denorm).to(txt_delta)
txt_delta = (txt_delta +
txt_delta_step).detach()
if opts.adv_max_norm > 0:
txt_delta = torch.clamp(
txt_delta, -opts.adv_max_norm,
opts.adv_max_norm).detach()
# (4) update and clip for image delta
if "image" in opts.adv_modality:
if opts.norm_type == "l2":
denorm = torch.norm(img_delta_grad.view(
img_delta_grad.size(0), -1),
dim=1).view(-1, 1, 1)
denorm = torch.clamp(denorm, min=1e-8)
img_delta_step = (opts.adv_lr_img *
img_delta_grad /
denorm).to(img_delta)
img_delta = (img_delta +
img_delta_step).detach()
if opts.adv_max_norm > 0:
delta_norm = torch.norm(img_delta.view(
img_delta.size(0), -1),
p=2,
dim=1).detach()
exceed_mask = (
delta_norm >
opts.adv_max_norm).to(img_embeds_init)
reweights = (opts.adv_max_norm /
delta_norm * exceed_mask +
(1 - exceed_mask)).view(
-1, 1, 1)
img_delta = (img_delta *
reweights).detach()
elif opts.norm_type == "linf":
denorm = torch.norm(img_delta_grad.view(
img_delta_grad.size(0), -1),
dim=1,
p=float("inf")).view(
-1, 1, 1)
denorm = torch.clamp(denorm, min=1e-8)
img_delta_step = (opts.adv_lr_img *
img_delta_grad /
denorm).to(img_delta)
img_delta = (img_delta +
img_delta_step).detach()
if opts.adv_max_norm > 0:
img_delta = torch.clamp(
img_delta, -opts.adv_max_norm,
opts.adv_max_norm).detach()
else:
loss = model(batch, compute_loss=True)
loss = loss.mean()
delay_unscale = (step +
1) % opts.gradient_accumulation_steps != 0
with amp.scale_loss(
loss, optimizer,
delay_unscale=delay_unscale) as scaled_loss:
scaled_loss.backward()
if not delay_unscale:
# gather gradients from every processes
# do this before unscaling to make sure every process uses
# the same gradient scale
grads = [
p.grad.data for p in model.parameters()
if p.requires_grad and p.grad is not None
]
all_reduce_and_rescale_tensors(grads, float(1))
running_loss(loss.item())
"""
loss compute end
log & step start
"""
if (step + 1) % opts.gradient_accumulation_steps == 0:
global_step += 1
# learning rate scheduling
lr_this_step = get_lr_sched(global_step, opts)
for i, param_group in enumerate(optimizer.param_groups):
if i == 0 or i == 1:
param_group['lr'] = lr_this_step * opts.lr_mul
elif i == 2 or i == 3:
param_group['lr'] = lr_this_step
else:
raise ValueError()
TB_LOGGER.add_scalar('lr', lr_this_step, global_step)
# log loss
# NOTE: not gathered across GPUs for efficiency
TB_LOGGER.add_scalar('loss', running_loss.val, global_step)
TB_LOGGER.step()
# update model params
if opts.grad_norm != -1:
grad_norm = clip_grad_norm_(
amp.master_params(optimizer), opts.grad_norm)
TB_LOGGER.add_scalar('grad_norm', grad_norm,
global_step)
optimizer.step()
optimizer.zero_grad()
pbar.update(1)
if global_step % 100 == 0:
# monitor training throughput
LOGGER.info(
f'============Step {global_step}=============')
tot_ex = sum(all_gather_list(n_examples))
ex_per_sec = int(tot_ex / (time() - start))
LOGGER.info(f'{tot_ex} examples trained at '
f'{ex_per_sec} ex/s')
TB_LOGGER.add_scalar('perf/ex_per_s', ex_per_sec,
global_step)
LOGGER.info(
f'===========================================')
if global_step % opts.valid_steps == 0:
val_log, results = validate(model, val_dataloader,
None)
with open(
f'{opts.output_dir}/results/'
f'results_{global_step}_'
f'rank{rank}.json', 'w') as f:
json.dump(results, f)
pd.DataFrame.from_dict(results).to_csv(
f'{opts.output_dir}/results/'
f'results_{global_step}_'
f'rank{rank}.csv',
index=False)
# _, test_results = test(model, test_dataloader, global_step)
# pd.DataFrame.from_dict(test_results).to_csv(
# f'{opts.output_dir}/results/'
# f'test_{global_step}.csv',
# index=False)
TB_LOGGER.log_scaler_dict(val_log)
model_saver.save(model, global_step)
if tuning:
with tune.checkpoint_dir(
step=n_epoch) as checkpoint_dir:
logger.info(
f'***** Save tune ckpt: {checkpoint_dir} *****'
)
path = os.path.join(checkpoint_dir,
"checkpoint")
torch.save((model.state_dict(),
optimizer.state_dict(), n_epoch,
n_examples), path)
tune.report(
loss=(val_log['valid/loss']),
accuracy=val_log['valid/acc'],
auroc=val_log['valid/auroc'],
)
if global_step >= opts.num_train_steps:
break
if global_step >= opts.num_train_steps:
break
n_epoch += 1
LOGGER.info(f"finished {n_epoch} epochs")
"""
END of training loop
"""
if opts.num_train_steps % opts.valid_steps != 0:
val_log, results = validate(model, val_dataloader, None)
with open(
f'{opts.output_dir}/results/'
f'results_{global_step}_'
f'rank{rank}.json', 'w') as f:
json.dump(results, f)
pd.DataFrame.from_dict(results).to_csv(
f'{opts.output_dir}/results/'
f'results_{global_step}_'
f'rank{rank}.csv',
index=False)
TB_LOGGER.log_scaler_dict(val_log)
model_saver.save(model, global_step)
def _gradient_accumulation(self, true_batchs, normalization, total_stats,
report_stats):
if self.grad_accum_count > 1:
self.model.zero_grad()
for batch in true_batchs:
target_size = batch.tgt.size(0)
# Truncated BPTT: reminder not compatible with accum > 1
if self.trunc_size:
trunc_size = self.trunc_size
else:
trunc_size = target_size
# dec_state = None
src = make_features(batch, 'src')#32*113
src = src.transpose(0, 1).contiguous()
# _, src_lengths = batch.src
#in here j ignored the length information and select a fix length
src_lengths=(torch.ones(batch.batch_size)*src.size(1)).long()
tgt_outer = make_features(batch, 'tgt')
for j in range(0, target_size-1, trunc_size):
# 1. Create truncated target.
tgt = tgt_outer[j: j + trunc_size]
# 2. F-prop all but generator.
if self.grad_accum_count == 1:
self.model.zero_grad()
outputs, attns = \
self.model(src, tgt, src_lengths)
# 3. Compute loss in shards for memory efficiency.
batch_stats = self.train_loss.sharded_compute_loss(
batch, outputs, attns, j,
trunc_size, self.shard_size, normalization)
total_stats.update(batch_stats)
report_stats.update(batch_stats)
# 4. Update the parameters and statistics.
if self.grad_accum_count == 1:
# Multi GPU gradient gather
if self.n_gpu > 1:
grads = [p.grad.data for p in self.model.parameters()
if p.requires_grad
and p.grad is not None]
all_reduce_and_rescale_tensors(
grads, float(1))
self.optim.step()
# If truncated, don't backprop fully.
# TO CHECK
# if dec_state is not None:
# dec_state.detach()
if self.model.decoder.state is not None:
self.model.decoder.detach_state()
# in case of multi step gradient accumulation,
# update only after accum batches
if self.grad_accum_count > 1:
if self.n_gpu > 1:
grads = [p.grad.data for p in self.model.parameters()
if p.requires_grad
and p.grad is not None]
all_reduce_and_rescale_tensors(
grads, float(1))
self.optim.step()
def main(opts):
hvd.init()
n_gpu = hvd.size()
device = torch.device("cuda", hvd.local_rank())
torch.cuda.set_device(hvd.local_rank())
opts.n_gpu = n_gpu
LOGGER.info("device: {} n_gpu: {}, rank: {}, "
"16-bits training: {}".format(device, n_gpu, hvd.rank(),
opts.fp16))
if hvd.rank() != 0:
LOGGER.disabled = True
set_random_seed(opts.seed)
# train_examples = None
LOGGER.info(f"Loading the whole video dataset {opts.sub_txt_db}, "
f"{opts.vfeat_db}")
video_db = load_video_sub_dataset(opts.vfeat_db, opts.sub_txt_db,
opts.vfeat_interval, opts)
# data loaders
# train
LOGGER.info(f"Loading the train QA dataset {opts.train_query_txt_db}")
video_ids = get_video_ids(opts.train_query_txt_db)
train_q_txt_db = QaQueryTokLmdb(opts.train_query_txt_db, opts.max_txt_len)
train_dataloaders = build_downstream_dataloaders([opts.task],
video_db,
video_ids,
True,
opts,
q_txt_db=train_q_txt_db,
shuffle=True)
meta_loader = MetaLoader(train_dataloaders,
accum_steps=opts.gradient_accumulation_steps,
distributed=n_gpu > 1)
meta_loader = PrefetchLoader(meta_loader)
# val
LOGGER.info(f"Loading the val QA dataset {opts.val_query_txt_db}")
video_ids = get_video_ids(opts.val_query_txt_db)
val_q_txt_db = QaQueryTokLmdb(opts.val_query_txt_db, -1)
val_dataloaders = build_downstream_dataloaders([opts.task],
video_db,
video_ids,
False,
opts,
q_txt_db=val_q_txt_db)
if opts.test_query_txt_db:
LOGGER.info(f"Loading the test QA dataset {opts.test_query_txt_db}")
video_ids = get_video_ids(opts.test_query_txt_db)
test_q_txt_db = QaQueryTokLmdb(opts.test_query_txt_db, -1)
test_dataloaders = build_downstream_dataloaders([opts.task],
video_db,
video_ids,
False,
opts,
q_txt_db=test_q_txt_db)
# Prepare model
if opts.checkpoint:
checkpoint = torch.load(opts.checkpoint)
else:
checkpoint = {}
img_pos_embed_weight_key = "v_encoder.f_encoder.img_embeddings" +\
".position_embeddings.weight"
if img_pos_embed_weight_key in checkpoint:
max_frm_seq_len = len(checkpoint[img_pos_embed_weight_key])
else:
max_frm_seq_len = MAX_FRM_SEQ_LEN
model = HeroForVideoQA.from_pretrained(opts.model_config,
state_dict=checkpoint,
vfeat_dim=VFEAT_DIM,
max_frm_seq_len=max_frm_seq_len)
model.to(device)
# make sure every process has same model parameters in the beginning
broadcast_tensors([p.data for p in model.parameters()], 0)
set_dropout(model, opts.dropout)
# Prepare optimizer
optimizer = build_optimizer(model, opts)
task2scaler = {t: i for i, t in enumerate(train_dataloaders.keys())}
model, optimizer = amp.initialize(model,
optimizer,
num_losses=len(task2scaler),
enabled=opts.fp16,
opt_level='O2')
restorer = TrainingRestorer(opts, model, optimizer)
global_step = restorer.global_step
TB_LOGGER.global_step = global_step
if hvd.rank() == 0:
save_training_meta(opts)
TB_LOGGER.create(join(opts.output_dir, 'log'))
pbar = tqdm(total=opts.num_train_steps)
model_saver = ModelSaver(join(opts.output_dir, 'ckpt'))
if not exists(join(opts.output_dir, 'results')):
# store tvqa predictions
os.makedirs(join(opts.output_dir, 'results'))
add_log_to_file(join(opts.output_dir, 'log', 'log.txt'))
else:
LOGGER.disabled = True
pbar = NoOp()
model_saver = NoOp()
restorer = NoOp()
if global_step > 0:
pbar.update(global_step)
LOGGER.info(f"***** Running training with {n_gpu} GPUs *****")
LOGGER.info(" Batch size = %d", opts.train_batch_size)
LOGGER.info(" Accumulate steps = %d", opts.gradient_accumulation_steps)
LOGGER.info(" Num steps = %d", opts.num_train_steps)
task2loss = {
task: RunningMeter(f'loss/{task}')
for task in train_dataloaders.keys()
}
for obj in (f'{opts.task}_qa', f'{opts.task}_st_ed'):
task2loss[obj] = RunningMeter(f'loss/{obj}')
model.train()
n_examples = defaultdict(int)
start = time()
# quick hack for amp delay_unscale bug
optimizer.zero_grad()
if global_step == 0:
optimizer.step()
for step, (task, batch) in enumerate(meta_loader):
n_examples[task] += opts.train_batch_size
loss = model(batch, task=task, compute_loss=True)
loss_qa, loss_st_ed = loss
loss = loss_qa + opts.lw_st_ed * loss_st_ed
for n, ls in (('st_ed', loss_st_ed), ('qa', loss_qa)):
ls = ls.item()
task2loss[f'{task}_{n}'](ls)
loss = loss.mean()
task2loss[task](loss.item())
delay_unscale = (step + 1) % opts.gradient_accumulation_steps != 0
with amp.scale_loss(loss,
optimizer,
delay_unscale=delay_unscale,
loss_id=task2scaler[task]) as scaled_loss:
scaled_loss.backward()
if not delay_unscale:
# gather gradients from every processes
# do this before unscaling to make sure every process uses
# the same gradient scale
grads = [
p.grad.data for p in model.parameters()
if p.requires_grad and p.grad is not None
]
all_reduce_and_rescale_tensors(grads, float(1))
if (step + 1) % opts.gradient_accumulation_steps == 0:
global_step += 1
# learning rate scheduling
lr_this_step = get_lr_sched(global_step, opts)
for i, param_group in enumerate(optimizer.param_groups):
if i == 0 or i == 1:
param_group['lr'] = lr_this_step * opts.lr_mul
elif i == 2 or i == 3:
param_group['lr'] = lr_this_step
else:
raise ValueError()
TB_LOGGER.add_scalar('lr', lr_this_step, global_step)
TB_LOGGER.log_scaler_dict({
temp_loss.name: temp_loss.val
for temp_loss in task2loss.values()
if temp_loss.val is not None
})
TB_LOGGER.step()
# update model params
if opts.grad_norm != -1:
grad_norm = clip_grad_norm_(amp.master_params(optimizer),
opts.grad_norm)
TB_LOGGER.add_scalar('grad_norm', grad_norm, global_step)
optimizer.step()
optimizer.zero_grad()
restorer.step()
pbar.update(1)
if global_step % 100 == 0:
# monitor training throughput
LOGGER.info('-------------------------------------------')
LOGGER.info(f'Step {global_step}:')
for t in train_dataloaders.keys():
tot_ex = sum(all_gather_list(n_examples[t]))
ex_per_sec = int(tot_ex / (time() - start))
LOGGER.info(f'{t}: {tot_ex} examples trained at '
f'{ex_per_sec} ex/s')
TB_LOGGER.add_scalar(f'perf/{t}_ex_per_s', ex_per_sec,
global_step)
if global_step % opts.valid_steps == 0:
LOGGER.info('===========================================')
LOGGER.info(f"Step {global_step}: start running validation")
validate(model,
val_dataloaders,
"val",
opts,
global_step=global_step)
if opts.test_query_txt_db:
validate(model,
test_dataloaders,
"test",
opts,
global_step=global_step)
LOGGER.info('===========================================')
model_saver.save(model, global_step)
if global_step >= opts.num_train_steps:
break
LOGGER.info('===========================================')
if global_step % opts.valid_steps != 0:
LOGGER.info('===========================================')
LOGGER.info(f"Step {global_step}: start running validation")
validate(model, val_dataloaders, "val", opts, global_step=global_step)
if opts.test_query_txt_db:
validate(model,
test_dataloaders,
"test",
opts,
global_step=global_step)
LOGGER.info('===========================================')
model_saver.save(model, f'{global_step}_final')
def main(opts):
hvd.init()
n_gpu = hvd.size()
device = torch.device("cuda", hvd.local_rank())
torch.cuda.set_device(hvd.local_rank())
rank = hvd.rank()
opts.rank = rank
LOGGER.info("device: {} n_gpu: {}, rank: {}, "
"16-bits training: {}".format(device, n_gpu, hvd.rank(),
opts.fp16))
if opts.gradient_accumulation_steps < 1:
raise ValueError("Invalid gradient_accumulation_steps parameter: {}, "
"should be >= 1".format(
opts.gradient_accumulation_steps))
set_random_seed(opts.seed)
# train_examples = None
LOGGER.info(f"Loading Train Dataset {opts.train_txt_db}, "
f"{opts.train_img_dir}")
if 'paired' in opts.model:
DatasetCls = Nlvr2PairedDataset
EvalDatasetCls = Nlvr2PairedEvalDataset
collate_fn = nlvr2_paired_collate
eval_collate_fn = nlvr2_paired_eval_collate
if opts.model == 'paired':
ModelCls = UniterForNlvr2Paired
elif opts.model == 'paired-attn':
ModelCls = UniterForNlvr2PairedAttn
else:
raise ValueError('unrecognized model type')
elif opts.model == 'triplet':
DatasetCls = Nlvr2TripletDataset
EvalDatasetCls = Nlvr2TripletEvalDataset
ModelCls = UniterForNlvr2Triplet
collate_fn = nlvr2_triplet_collate
eval_collate_fn = nlvr2_triplet_eval_collate
else:
raise ValueError('unrecognized model type')
# data loaders
train_dataloader = create_dataloader(opts.train_img_db, opts.train_txt_db,
opts.train_batch_size, True,
DatasetCls, collate_fn, opts)
val_dataloader = create_dataloader(opts.val_img_db, opts.val_txt_db,
opts.val_batch_size, False,
EvalDatasetCls, eval_collate_fn, opts)
test_dataloader = create_dataloader(opts.test_img_db, opts.test_txt_db,
opts.val_batch_size, False,
EvalDatasetCls, eval_collate_fn, opts)
# Prepare model
if opts.checkpoint:
checkpoint = torch.load(opts.checkpoint)
else:
checkpoint = {}
model = ModelCls.from_pretrained(opts.model_config,
state_dict=checkpoint,
img_dim=IMG_DIM)
model.init_type_embedding()
model.to(device)
# make sure every process has same model parameters in the beginning
broadcast_tensors([p.data for p in model.parameters()], 0)
set_dropout(model, opts.dropout)
# Prepare optimizer
optimizer = build_optimizer(model, opts)
model, optimizer = amp.initialize(model,
optimizer,
enabled=opts.fp16,
opt_level='O2')
global_step = 0
if rank == 0:
save_training_meta(opts)
TB_LOGGER.create(join(opts.output_dir, 'log'))
pbar = tqdm(total=opts.num_train_steps)
model_saver = ModelSaver(join(opts.output_dir, 'ckpt'))
os.makedirs(join(opts.output_dir, 'results')) # store val predictions
add_log_to_file(join(opts.output_dir, 'log', 'log.txt'))
else:
LOGGER.disabled = True
pbar = NoOp()
model_saver = NoOp()
LOGGER.info(f"***** Running training with {n_gpu} GPUs *****")
LOGGER.info(" Num examples = %d", len(train_dataloader.dataset))
LOGGER.info(" Batch size = %d", opts.train_batch_size)
LOGGER.info(" Accumulate steps = %d", opts.gradient_accumulation_steps)
LOGGER.info(" Num steps = %d", opts.num_train_steps)
running_loss = RunningMeter('loss')
model.train()
n_examples = 0
n_epoch = 0
start = time()
# quick hack for amp delay_unscale bug
optimizer.zero_grad()
optimizer.step()
while True:
for step, batch in enumerate(train_dataloader):
targets = batch['targets']
n_examples += targets.size(0)
loss = model(**batch, compute_loss=True)
loss = loss.mean()
delay_unscale = (step + 1) % opts.gradient_accumulation_steps != 0
with amp.scale_loss(loss, optimizer,
delay_unscale=delay_unscale) as scaled_loss:
scaled_loss.backward()
if not delay_unscale:
# gather gradients from every processes
# do this before unscaling to make sure every process uses
# the same gradient scale
grads = [
p.grad.data for p in model.parameters()
if p.requires_grad and p.grad is not None
]
all_reduce_and_rescale_tensors(grads, float(1))
running_loss(loss.item())
if (step + 1) % opts.gradient_accumulation_steps == 0:
global_step += 1
# learning rate scheduling
lr_this_step = get_lr_sched(global_step, opts)
for param_group in optimizer.param_groups:
param_group['lr'] = lr_this_step
TB_LOGGER.add_scalar('lr', lr_this_step, global_step)
# log loss
losses = all_gather_list(running_loss)
running_loss = RunningMeter(
'loss',
sum(l.val for l in losses) / len(losses))
TB_LOGGER.add_scalar('loss', running_loss.val, global_step)
TB_LOGGER.step()
# update model params
if opts.grad_norm != -1:
grad_norm = clip_grad_norm_(amp.master_params(optimizer),
opts.grad_norm)
TB_LOGGER.add_scalar('grad_norm', grad_norm, global_step)
optimizer.step()
optimizer.zero_grad()
pbar.update(1)
if global_step % 100 == 0:
# monitor training throughput
tot_ex = sum(all_gather_list(n_examples))
ex_per_sec = int(tot_ex / (time() - start))
LOGGER.info(f'Step {global_step}: '
f'{tot_ex} examples trained at '
f'{ex_per_sec} ex/s')
TB_LOGGER.add_scalar('perf/ex_per_s', ex_per_sec,
global_step)
if global_step % opts.valid_steps == 0:
for split, loader in [('val', val_dataloader),
('test', test_dataloader)]:
LOGGER.info(f"Step {global_step}: start running "
f"validation on {split} split...")
log, results = validate(model, loader, split)
with open(
f'{opts.output_dir}/results/'
f'{split}_results_{global_step}_'
f'rank{rank}.csv', 'w') as f:
for id_, ans in results:
f.write(f'{id_},{ans}\n')
TB_LOGGER.log_scaler_dict(log)
model_saver.save(model, global_step)
if global_step >= opts.num_train_steps:
break
if global_step >= opts.num_train_steps:
break
n_epoch += 1
LOGGER.info(f"Step {global_step}: finished {n_epoch} epochs")
for split, loader in [('val', val_dataloader), ('test', test_dataloader)]:
LOGGER.info(f"Step {global_step}: start running "
f"validation on {split} split...")
log, results = validate(model, loader, split)
with open(
f'{opts.output_dir}/results/'
f'{split}_results_{global_step}_'
f'rank{rank}_final.csv', 'w') as f:
for id_, ans in results:
f.write(f'{id_},{ans}\n')
TB_LOGGER.log_scaler_dict(log)
model_saver.save(model, f'{global_step}_final')
def main(opts):
hvd.init()
n_gpu = hvd.size()
device = torch.device("cuda", hvd.local_rank())
torch.cuda.set_device(hvd.local_rank())
rank = hvd.rank()
opts.rank = rank
LOGGER.info("device: {} n_gpu: {}, rank: {}, "
"16-bits training: {}".format(device, n_gpu, hvd.rank(),
opts.fp16))
if opts.gradient_accumulation_steps < 1:
raise ValueError("Invalid gradient_accumulation_steps parameter: {}, "
"should be >= 1".format(
opts.gradient_accumulation_steps))
set_random_seed(opts.seed)
# load DBs and image dirs
all_img_dbs = ImageLmdbGroup(opts.conf_th, opts.max_bb, opts.min_bb,
opts.num_bb, opts.compressed_db)
# train
LOGGER.info(f"Loading Train Dataset "
f"{opts.train_txt_dbs}, {opts.train_img_dbs}")
train_datasets = []
for txt_path, img_path in zip(opts.train_txt_dbs, opts.train_img_dbs):
img_db, img_db_gt = load_img_feat(img_path, all_img_dbs, opts)
qa_txt_db = VcrTxtTokLmdb(txt_path, opts.max_txt_len, task="qa")
qar_txt_db = VcrTxtTokLmdb(txt_path, opts.max_txt_len, task="qar")
train_datasets.append(
VcrDataset(qa_txt_db, img_db_gt=img_db_gt, img_db=img_db))
train_datasets.append(
VcrDataset(qar_txt_db, img_db_gt=img_db_gt, img_db=img_db))
train_dataset = ConcatDatasetWithLens(train_datasets)
train_dataloader = build_dataloader(train_dataset, vcr_collate, True, opts)
# val
LOGGER.info(f"Loading Val Dataset {opts.val_txt_db}, {opts.val_img_db}")
val_img_db, val_img_db_gt = load_img_feat(opts.val_img_db, all_img_dbs,
opts)
val_txt_db = VcrTxtTokLmdb(opts.val_txt_db, -1)
val_dataset = VcrEvalDataset("val",
val_txt_db,
img_db=val_img_db,
img_db_gt=val_img_db_gt)
val_final_dataset = VcrEvalDataset("test",
val_txt_db,
img_db=val_img_db,
img_db_gt=val_img_db_gt)
val_dataloader = build_dataloader(val_dataset, vcr_eval_collate, False,
opts)
val_final_dataloader = build_dataloader(val_final_dataset,
vcr_eval_collate, False, opts)
# Prepare model
if opts.checkpoint and opts.checkpoint_from == "pretrain":
ckpt = torch.load(opts.checkpoint)
checkpoint = {k.replace('bert', 'uniter'): v for k, v in ckpt.items()}
else:
checkpoint = {}
all_dbs = opts.train_txt_dbs + [opts.val_txt_db]
toker = json.load(open(f'{all_dbs[0]}/meta.json'))['bert']
assert all(toker == json.load(open(f'{db}/meta.json'))['bert']
for db in all_dbs)
model = UniterForVisualCommonsenseReasoning.from_pretrained(
opts.model_config, checkpoint, img_dim=IMG_DIM)
model.init_type_embedding()
model.init_word_embedding(NUM_SPECIAL_TOKENS)
if opts.checkpoint_from == "vcr_pretrain":
ckpt = torch.load(opts.checkpoint)
checkpoint = {k.replace('bert', 'uniter'): v for k, v in ckpt.items()}
state_dict = checkpoint.get('model_state', checkpoint)
matched_state_dict = {}
unexpected_keys = set()
missing_keys = set()
for name, param in model.named_parameters():
missing_keys.add(name)
for key, data in state_dict.items():
if key in missing_keys:
matched_state_dict[key] = data
missing_keys.remove(key)
else:
unexpected_keys.add(key)
print("Unexpected_keys:", list(unexpected_keys))
print("Missing_keys:", list(missing_keys))
model.load_state_dict(matched_state_dict, strict=False)
del checkpoint
model.to(device)
# make sure every process has same model parameters in the beginning
broadcast_tensors([p.data for p in model.parameters()], 0)
set_dropout(model, opts.dropout)
# Prepare optimizer
optimizer = build_optimizer(model, opts)
model, optimizer = amp.initialize(model,
optimizer,
enabled=opts.fp16,
opt_level='O2')
global_step = 0
if rank == 0:
save_training_meta(opts)
TB_LOGGER.create(join(opts.output_dir, 'log'))
pbar = tqdm(total=opts.num_train_steps)
model_saver = ModelSaver(join(opts.output_dir, 'ckpt'))
os.makedirs(join(opts.output_dir, 'results')) # store VQA predictions
add_log_to_file(join(opts.output_dir, 'log', 'log.txt'))
else:
LOGGER.disabled = True
pbar = NoOp()
model_saver = NoOp()
LOGGER.info(f"***** Running training with {n_gpu} GPUs *****")
LOGGER.info(" Num examples = %d", len(train_dataset) * hvd.size())
LOGGER.info(" Batch size = %d", opts.train_batch_size)
LOGGER.info(" Accumulate steps = %d", opts.gradient_accumulation_steps)
LOGGER.info(" Num steps = %d", opts.num_train_steps)
running_loss = RunningMeter('loss')
model.train()
n_examples = 0
n_epoch = 0
start = time()
# quick hack for amp delay_unscale bug
optimizer.zero_grad()
optimizer.step()
while True:
for step, batch in enumerate(train_dataloader):
n_examples += batch['input_ids'].size(0)
# ============= Code for adversarial training =============
if opts.adv_training:
# initialize delta
txt_embeds_init = model.uniter.embeddings.word_embeddings(
batch['input_ids'])
img_embeds_init = batch['img_feat']
# for simplicity, we initialize the delta as zero vectors, which performs
# very simliar as initializing randomly using norm or uniform distributions
txt_delta = torch.zeros_like(txt_embeds_init)
img_delta = torch.zeros_like(img_embeds_init)
# calculate the prob. scores for clean samples
gt_answer_scores = model(batch, compute_loss=False)
gt_answer_prob = F.softmax(gt_answer_scores, dim=1)
gt_answer_logprob = F.log_softmax(gt_answer_scores, dim=1)
# the main loop
for astep in range(opts.adv_steps):
# (0) forward
if opts.adv_modality == ["text"]:
txt_delta.requires_grad_()
img_delta = torch.zeros_like(img_embeds_init)
elif opts.adv_modality == ["image"]:
img_delta.requires_grad_()
txt_delta = torch.zeros_like(txt_embeds_init)
else:
txt_delta.requires_grad_()
img_delta.requires_grad_()
if "alter" not in opts.adv_modality:
answer_scores = model(batch,
adv_training=True,
adv_modality=opts.adv_modality,
adv_delta_txt=txt_delta,
adv_delta_img=img_delta,
compute_loss=False)
# CE loss
ce_loss = F.cross_entropy(answer_scores,
batch['targets'].squeeze(-1),
reduction='mean')
# KL loss
answer_prob = F.softmax(answer_scores, dim=1)
answer_logprob = F.log_softmax(answer_scores, dim=1)
kl_loss = F.kl_div(
answer_logprob, gt_answer_prob, reduction='none') + \
F.kl_div(
gt_answer_logprob, answer_prob,
reduction='none')
kl_loss = kl_loss.mean()
# (1) backward
loss = (ce_loss +
opts.adv_kl_weight * kl_loss) / opts.adv_steps
else:
answer_scores_1 = model(batch,
adv_training=True,
adv_modality=["text"],
adv_delta_txt=txt_delta,
adv_delta_img=None,
compute_loss=False)
# CE loss
ce_loss_1 = F.cross_entropy(
answer_scores,
batch['targets'].squeeze(-1),
reduction='mean')
answer_scores_2 = model(batch,
adv_training=True,
adv_modality=["image"],
adv_delta_txt=None,
adv_delta_img=img_delta,
compute_loss=False)
# CE loss
ce_loss_2 = F.cross_entropy(
answer_scores,
batch['targets'].squeeze(-1),
reduction='mean')
# KL loss
answer_prob_1 = F.softmax(answer_scores_1, dim=1)
answer_logprob_1 = F.log_softmax(answer_scores_1,
dim=1)
answer_prob_2 = F.softmax(answer_scores_2, dim=1)
answer_logprob_2 = F.log_softmax(answer_scores_2,
dim=1)
kl_loss_1 = F.kl_div(
answer_logprob_1, gt_answer_prob, reduction='none') + \
F.kl_div(
gt_answer_logprob, answer_prob_1,
reduction='none')
kl_loss_1 = kl_loss_1.mean()
kl_loss_2 = F.kl_div(
answer_logprob_2, gt_answer_prob, reduction='none') + \
F.kl_div(
gt_answer_logprob, answer_prob_2,
reduction='none')
kl_loss_2 = kl_loss_2.mean()
# (1) backward
loss = (ce_loss_1 + ce_loss_2 + opts.adv_kl_weight *
(kl_loss_1 + kl_loss_2)) / (opts.adv_steps * 2)
delay_unscale = (
(step + 1) % opts.gradient_accumulation_steps !=
0) or ((astep + 1) % opts.adv_steps != 0)
with amp.scale_loss(
loss, optimizer,
delay_unscale=delay_unscale) as scaled_loss:
scaled_loss.backward(retain_graph=True)
if not delay_unscale:
# gather gradients from every processes
# do this before unscaling
# to make sure every process uses
# the same gradient scale
grads = [
p.grad.data for p in model.parameters()
if p.requires_grad and p.grad is not None
]
all_reduce_and_rescale_tensors(grads, float(1))
running_loss(loss.item())
if astep == opts.adv_steps - 1:
# further updates on delta
break
# (2) get gradient on delta
# fix fp16 problem
amp_scale = scaled_loss.item() // loss.item()
if "text" in opts.adv_modality:
txt_delta_grad = txt_delta.grad.clone().detach()
txt_delta_grad = txt_delta_grad.float() / amp_scale
if "image" in opts.adv_modality:
img_delta_grad = img_delta.grad.clone().detach()
img_delta_grad = img_delta_grad.float() / amp_scale
# (3) update and clip for txt delta
if "text" in opts.adv_modality:
if opts.norm_type == "l2":
denorm = torch.norm(txt_delta_grad.view(
txt_delta_grad.size(0), -1),
dim=1).view(-1, 1, 1)
denorm = torch.clamp(denorm, min=1e-8)
txt_delta_step = (opts.adv_lr_txt *
txt_delta_grad /
denorm).to(txt_delta)
txt_delta = (txt_delta + txt_delta_step).detach()
if opts.adv_max_norm > 0:
delta_norm = torch.norm(txt_delta.view(
txt_delta.size(0), -1),
p=2,
dim=1).detach()
exceed_mask = (delta_norm > opts.adv_max_norm
).to(txt_embeds_init)
reweights = (opts.adv_max_norm / delta_norm *
exceed_mask +
(1 - exceed_mask)).view(-1, 1, 1)
txt_delta = (txt_delta * reweights).detach()
elif opts.norm_type == "linf":
denorm = torch.norm(txt_delta_grad.view(
txt_delta_grad.size(0), -1),
dim=1,
p=float("inf")).view(-1, 1, 1)
denorm = torch.clamp(denorm, min=1e-8)
txt_delta_step = (opts.adv_lr_txt *
txt_delta_grad /
denorm).to(txt_delta)
txt_delta = (txt_delta + txt_delta_step).detach()
if opts.adv_max_norm > 0:
txt_delta = torch.clamp(
txt_delta, -opts.adv_max_norm,
opts.adv_max_norm).detach()
# (4) update and clip for image delta
if "image" in opts.adv_modality:
if opts.norm_type == "l2":
denorm = torch.norm(img_delta_grad.view(
img_delta_grad.size(0), -1),
dim=1).view(-1, 1, 1)
denorm = torch.clamp(denorm, min=1e-8)
img_delta_step = (opts.adv_lr_img *
img_delta_grad /
denorm).to(img_delta)
img_delta = (img_delta + img_delta_step).detach()
if opts.adv_max_norm > 0:
delta_norm = torch.norm(img_delta.view(
img_delta.size(0), -1),
p=2,
dim=1).detach()
exceed_mask = (delta_norm > opts.adv_max_norm
).to(img_embeds_init)
reweights = (opts.adv_max_norm / delta_norm *
exceed_mask +
(1 - exceed_mask)).view(-1, 1, 1)
img_delta = (img_delta * reweights).detach()
elif opts.norm_type == "linf":
denorm = torch.norm(img_delta_grad.view(
img_delta_grad.size(0), -1),
dim=1,
p=float("inf")).view(-1, 1, 1)
denorm = torch.clamp(denorm, min=1e-8)
img_delta_step = (opts.adv_lr_img *
img_delta_grad /
denorm).to(img_delta)
img_delta = (img_delta + img_delta_step).detach()
if opts.adv_max_norm > 0:
img_delta = torch.clamp(
img_delta, -opts.adv_max_norm,
opts.adv_max_norm).detach()
else:
loss = model(batch, compute_loss=True)
loss = loss.mean()
delay_unscale = ((step + 1) % opts.gradient_accumulation_steps
!= 0)
with amp.scale_loss(
loss, optimizer,
delay_unscale=delay_unscale) as scaled_loss:
scaled_loss.backward()
if not delay_unscale:
# gather gradients from every processes
# do this before unscaling to make sure every process uses
# the same gradient scale
grads = [
p.grad.data for p in model.parameters()
if p.requires_grad and p.grad is not None
]
all_reduce_and_rescale_tensors(grads, float(1))
running_loss(loss.item())
# ============================ End ==========================
if (step + 1) % opts.gradient_accumulation_steps == 0:
global_step += 1
# learning rate scheduling
lr_this_step = get_lr_sched(global_step, opts)
for i, param_group in enumerate(optimizer.param_groups):
if i == 0 or i == 1:
param_group['lr'] = lr_this_step * opts.lr_mul
elif i == 2 or i == 3:
param_group['lr'] = lr_this_step
else:
raise ValueError()
TB_LOGGER.add_scalar('lr', lr_this_step, global_step)
# log loss
# NOTE: not gathered across GPUs for efficiency
TB_LOGGER.add_scalar('loss', running_loss.val, global_step)
TB_LOGGER.step()
# update model params
if opts.grad_norm != -1:
grad_norm = clip_grad_norm_(amp.master_params(optimizer),
opts.grad_norm)
TB_LOGGER.add_scalar('grad_norm', grad_norm, global_step)
optimizer.step()
optimizer.zero_grad()
pbar.update(1)
if global_step % 100 == 0:
# monitor training throughput
LOGGER.info(f'============Step {global_step}=============')
tot_ex = sum(all_gather_list(n_examples))
ex_per_sec = int(tot_ex / (time() - start))
LOGGER.info(f'{tot_ex} examples trained at '
f'{ex_per_sec} ex/s')
TB_LOGGER.add_scalar('perf/ex_per_s', ex_per_sec,
global_step)
LOGGER.info('===========================================')
if global_step % opts.valid_steps == 0:
val_log, results = validate(model, val_dataloader)
TB_LOGGER.log_scaler_dict(val_log)
model_saver.save(model, global_step)
if global_step >= opts.num_train_steps:
break
if global_step >= opts.num_train_steps:
break
n_epoch += 1
LOGGER.info(f"finished {n_epoch} epochs")
if global_step % opts.valid_steps != 0:
val_log, results = validate(model, val_dataloader)
TB_LOGGER.log_scaler_dict(val_log)
val_log, results = validate(model, val_final_dataloader)
with open(
f'{opts.output_dir}/results/'
f'results_{global_step}_final_qa_qar_'
f'rank{rank}.json', 'w') as f:
json.dump(results, f)
TB_LOGGER.log_scaler_dict(val_log)
model_saver.save(model, global_step)
def main(opts):
hvd.init()
n_gpu = hvd.size()
device = torch.device("cuda", hvd.local_rank())
torch.cuda.set_device(hvd.local_rank())
rank = hvd.rank()
opts.rank = rank
LOGGER.info("device: {} n_gpu: {}, rank: {}, "
"16-bits training: {}".format(
device, n_gpu, hvd.rank(), opts.fp16))
device = torch.device("cuda:1")
if opts.gradient_accumulation_steps < 1:
raise ValueError("Invalid gradient_accumulation_steps parameter: {}, "
"should be >= 1".format(
opts.gradient_accumulation_steps))
set_random_seed(opts.seed)
if hvd.rank() == 0:
TB_LOGGER.create(join(opts.output_dir, 'log'))
os.makedirs(join(opts.output_dir, 'ckpt'))
save_training_meta(opts)
# TB_LOGGER.create(join(opts.output_dir, 'log'))
model_saver = ModelSaver(join(opts.output_dir, 'ckpt'))
add_log_to_file(join(opts.output_dir, 'log', 'log.txt'))
# store ITM predictions
os.makedirs(join(opts.output_dir, 'results_val'))
os.makedirs(join(opts.output_dir, 'results_test'))
os.makedirs(join(opts.output_dir, 'results_train'))
else:
LOGGER.disabled = True
model_saver = NoOp()
# load DBs and image dirs
all_img_dbs = ImageLmdbGroup(opts.conf_th, opts.max_bb, opts.min_bb,
opts.num_bb, opts.compressed_db)
# train
train_dataset = MemeAIDataset(json_path = '/home/data/meme_json/train.json',
npz_folder = '/home/data/faster_cnn_feature/',
mode = 'train')
train_loader = DataLoader(train_dataset,
batch_size = opts.train_batch_size,
shuffle = True,
num_workers = opts.n_workers,
collate_fn=collate_fn)
train_loader = PrefetchLoader(train_loader)
# val
val_dataset = MemeAIDataset(json_path = '/home/data/meme_json/dev.json',
npz_folder = '/home/data/faster_cnn_feature/',
mode = 'val')
val_loader = DataLoader(val_dataset,
batch_size = opts.inf_minibatch_size,
shuffle = False,
num_workers = opts.n_workers,
collate_fn=collate_fn)
val_loader = PrefetchLoader(val_loader)
# Prepare model
if opts.checkpoint:
checkpoint = torch.load(opts.checkpoint)
else:
checkpoint = {}
model = Meme.from_pretrained(
opts.model_config, state_dict=checkpoint,
img_dim=IMG_DIM)
model.init_output() # pretrain ITM head is different from ranking head
model.to(device)
# make sure every process has same model parameters in the beginning
# broadcast_tensors([p.data for p in model.parameters()], 0)
# set_dropout(model, opts.dropout)
# Prepare optimizer
optimizer = build_optimizer(model, opts)
model, optimizer = amp.initialize(model, optimizer,
enabled=opts.fp16, opt_level='O2')
global_step = 0
# LOGGER.info(f"***** Running training on {n_gpu} GPUs *****")
# LOGGER.info(" Num examples = %d", len(train_dataset) * hvd.size())
LOGGER.info(" Batch size = %d", opts.train_batch_size)
LOGGER.info(" Accumulate steps = %d", opts.gradient_accumulation_steps)
LOGGER.info(" Num steps = %d", opts.num_train_steps)
running_loss = RunningMeter('loss')
model.train()
n_examples = 0
n_epoch = 0
start = time()
# quick hack for amp delay_unscale bug
optimizer.zero_grad()
optimizer.step()
# while True:
for epoch in range(opts.epoch):
print('epoch {}/ {}'.format(epoch, opts.epoch))
pbar = tqdm(total=len(train_loader))
model.train()
preds = None
gt = None
for step, batch in enumerate(train_loader):
x = batch[0]
y = batch[1]
n_examples += x['input_ids'].size(0)
pred = model(x)
if preds is None:
preds = torch.sigmoid(pred)
gt = y
else:
preds = torch.cat((preds, torch.sigmoid(pred)), dim = 0)
gt = torch.cat((gt, y), dim = 0)
loss = F.binary_cross_entropy(torch.sigmoid(pred), y)
delay_unscale = (step+1) % opts.gradient_accumulation_steps != 0
with amp.scale_loss(loss, optimizer, delay_unscale=delay_unscale
) as scaled_loss:
scaled_loss.backward()
if not delay_unscale:
# gather gradients from every processes
# do this before unscaling to make sure every process uses
# the same gradient scale
grads = [p.grad.data for p in model.parameters()
if p.requires_grad and p.grad is not None]
all_reduce_and_rescale_tensors(grads, float(1))
running_loss(loss.item())
if (step + 1) % opts.gradient_accumulation_steps == 0:
global_step += 1
# learning rate scheduling
lr_this_step = get_lr_sched(global_step, opts)
for param_group in optimizer.param_groups:
param_group['lr'] = lr_this_step
TB_LOGGER.add_scalar('lr', lr_this_step, global_step)
# log loss
# NOTE: not gathered across GPUs for efficiency
TB_LOGGER.add_scalar('loss', running_loss.val, global_step)
TB_LOGGER.step()
# update model params
if opts.grad_norm != -1:
grad_norm = clip_grad_norm_(amp.master_params(optimizer),
opts.grad_norm)
TB_LOGGER.add_scalar('grad_norm', grad_norm, global_step)
optimizer.step()
optimizer.zero_grad()
global_step += 1
# learning rate scheduling
lr_this_step = get_lr_sched(global_step, opts)
for param_group in optimizer.param_groups:
param_group['lr'] = lr_this_step
TB_LOGGER.add_scalar('lr', lr_this_step, global_step)
# log loss
# NOTE: not gathered across GPUs for efficiency
TB_LOGGER.add_scalar('loss', running_loss.val, global_step)
TB_LOGGER.step()
# update model params
if opts.grad_norm != -1:
grad_norm = clip_grad_norm_(amp.master_params(optimizer),
opts.grad_norm)
TB_LOGGER.add_scalar('grad_norm', grad_norm, global_step)
optimizer.step()
optimizer.zero_grad()
with torch.no_grad():
preds = preds.cpu().numpy().reshape(len(preds), )
gt = gt.cpu().numpy()
roc = roc_auc_score(gt, preds)
acc = accuracy_score(gt, np.around(preds))
train_log = {'train/roc': roc, 'train/acc': acc}
TB_LOGGER.log_scaler_dict({f"train/{k}": v for k, v in train_log.items()})
# monitor training throughput
val_log = validate(model, val_loader)
TB_LOGGER.log_scaler_dict({f"valid/{k}": v for k, v in val_log.items()})
LOGGER.info(train_log)
LOGGER.info(val_log)
model_saver.save(model, global_step)
pbar.close()
def main(opts):
hvd.init()
n_gpu = hvd.size()
device = torch.device("cuda", hvd.local_rank())
torch.cuda.set_device(hvd.local_rank())
opts.n_gpu = n_gpu
LOGGER.info("device: {} n_gpu: {}, rank: {}, "
"16-bits training: {}".format(device, n_gpu, hvd.rank(),
opts.fp16))
if hvd.rank() != 0:
LOGGER.disabled = True
set_random_seed(opts.seed)
# data loaders
train_dataloaders = {}
val_dataloaders = {}
for target, t_r in zip(opts.targets, opts.targets_ratio):
train_loaders, val_loaders = build_target_loaders(
target, t_r,
opts) # -> choose which task and get corrsponding task dataloder
train_dataloaders.update(train_loaders)
val_dataloaders.update(val_loaders)
meta_loader = MetaLoader(train_dataloaders,
accum_steps=opts.gradient_accumulation_steps,
distributed=n_gpu > 1)
meta_loader = PrefetchLoader(meta_loader)
# Prepare model
if opts.checkpoint:
checkpoint = torch.load(opts.checkpoint)
else:
checkpoint = {}
img_pos_embed_weight_key = "v_encoder.f_encoder.img_embeddings" +\
".position_embeddings.weight"
if img_pos_embed_weight_key in checkpoint:
max_frm_seq_len = len(checkpoint[img_pos_embed_weight_key])
else:
max_frm_seq_len = MAX_FRM_SEQ_LEN
if opts.load_partial_pretrained:
# from roberta
model = HeroForPretraining(VideoModelConfig(opts.model_config),
vfeat_dim=VFEAT_DIM,
max_frm_seq_len=max_frm_seq_len,
lw_neg_ctx=opts.lw_neg_ctx,
lw_neg_q=opts.lw_neg_q,
lw_st_ed=0,
ranking_loss_type=opts.ranking_loss_type,
use_hard_negative=False,
hard_pool_size=opts.hard_pool_size,
margin=opts.margin,
use_all_neg=opts.use_all_neg,
drop_svmr_prob=opts.drop_svmr_prob)
model.load_partial_pretrained(checkpoint,
VFEAT_DIM,
max_frm_seq_len,
skip_layers=opts.skip_layer_loading)
else:
# continue training
model = HeroForPretraining.from_pretrained(
opts.model_config,
state_dict=checkpoint,
vfeat_dim=VFEAT_DIM,
max_frm_seq_len=max_frm_seq_len,
lw_neg_ctx=opts.lw_neg_ctx,
lw_neg_q=opts.lw_neg_q,
lw_st_ed=0,
ranking_loss_type=opts.ranking_loss_type,
use_hard_negative=False,
hard_pool_size=opts.hard_pool_size,
margin=opts.margin,
use_all_neg=opts.use_all_neg,
drop_svmr_prob=opts.drop_svmr_prob)
model.to(device)
# make sure every process has same model parameters in the beginning
broadcast_tensors([p.data for p in model.parameters()], 0)
set_dropout(model, opts.dropout)
# Prepare optimizer
optimizer = build_optimizer(model, opts)
task2scaler = {t: i for i, t in enumerate(train_dataloaders.keys())}
model, optimizer = amp.initialize(model,
optimizer,
num_losses=len(task2scaler),
enabled=opts.fp16,
opt_level='O2')
restorer = TrainingRestorer(opts, model, optimizer)
all_gather_list(None) # sync to prevent slower rank to read training meta
global_step = restorer.global_step
TB_LOGGER.global_step = global_step
if hvd.rank() == 0:
save_training_meta(opts)
TB_LOGGER.create(join(opts.output_dir, 'log'))
pbar = tqdm(total=opts.num_train_steps)
model_saver = ModelSaver(join(opts.output_dir, 'ckpt'))
add_log_to_file(join(opts.output_dir, 'log', 'log.txt'))
else:
pbar = NoOp()
model_saver = NoOp()
restorer = NoOp()
if global_step > 0:
pbar.update(global_step)
LOGGER.info(f"***** Running training with {n_gpu} GPUs *****")
LOGGER.info(" Batch size = %d", opts.train_batch_size)
LOGGER.info(" Accumulate steps = %d", opts.gradient_accumulation_steps)
LOGGER.info(" Num steps = %d", opts.num_train_steps)
task2loss = {
task: RunningMeter(f'loss/{task}')
for task in train_dataloaders.keys()
}
for task in train_dataloaders.keys():
if task.startswith('vsm'):
for obj in ('st_ed', 'neg_ctx', 'neg_q'):
task2loss[f"{task}_{obj}"] = RunningMeter(f'loss/{task}_{obj}')
model.train()
n_examples = defaultdict(int)
start = time()
# quick hack for amp delay_unscale bug
optimizer.zero_grad()
if global_step == 0:
optimizer.step()
assert all(global_step == s for s in all_gather_list(global_step))
for step, (task, batch) in enumerate(meta_loader):
LOGGER.debug(f"Task: {task}")
# hard negative in VSM
if len(opts.hard_negtiave_start_step) > 0:
for i, hn_step in enumerate(opts.hard_negtiave_start_step):
if global_step >= hn_step and hn_step != -1:
model.set_hard_negative(True, opts.hard_pool_size[i],
opts.hard_neg_weights[i])
# start-end loss
if opts.train_span_start_step != -1 and\
global_step >= opts.train_span_start_step:
model.set_train_st_ed(opts.lw_st_ed)
train_task = task.split('_')[0]
n_examples[task] += opts.train_batch_size
loss = model(batch, task=train_task, compute_loss=True)
if train_task == 'vsm':
loss_st_ed, loss_neg_ctx, loss_neg_q = loss
loss = loss_st_ed + loss_neg_ctx + loss_neg_q
for n, ls, w in (('st_ed', loss_st_ed, opts.lw_st_ed),
('neg_ctx', loss_neg_ctx, opts.lw_neg_ctx),
('neg_q', loss_neg_q, opts.lw_neg_q)):
ls = ls.item()
if w:
ls /= w
task2loss[f'{task}_{n}'](ls)
elif train_task == "mffr":
loss = torch.sqrt(loss.sum(dim=1))
loss = loss.mean()
task2loss[task](loss.item())
delay_unscale = (step + 1) % opts.gradient_accumulation_steps != 0
with amp.scale_loss(loss,
optimizer,
delay_unscale=delay_unscale,
loss_id=task2scaler[task]) as scaled_loss:
scaled_loss.backward()
if not delay_unscale:
# gather gradients from every processes
# do this before unscaling to make sure every process uses
# the same gradient scale
grads = [
p.grad.data for p in model.parameters()
if p.requires_grad and p.grad is not None
]
LOGGER.debug("before reduce grad")
all_reduce_and_rescale_tensors(grads, float(1))
LOGGER.debug("after reduce grad")
if (step + 1) % opts.gradient_accumulation_steps == 0:
global_step += 1
# learning rate scheduling
lr_this_step = get_lr_sched(global_step, opts)
for param_group in optimizer.param_groups:
param_group['lr'] = lr_this_step
TB_LOGGER.add_scalar('lr', lr_this_step, global_step)
# log loss
# NOTE: only consider rank 0 for speed
TB_LOGGER.log_scaler_dict({
ll.name: ll.val
for ll in task2loss.values() if ll.val is not None
})
TB_LOGGER.step()
LOGGER.debug("before norm grad")
# update model params
if opts.grad_norm != -1:
grad_norm = clip_grad_norm_(amp.master_params(optimizer),
opts.grad_norm)
TB_LOGGER.add_scalar('grad_norm', grad_norm, global_step)
LOGGER.debug("after norm grad")
LOGGER.debug("before optim step")
optimizer.step()
optimizer.zero_grad()
pbar.update(1)
LOGGER.debug("after optim step")
if global_step % 100 == 0:
LOGGER.debug("after gather stats")
# monitor training throughput
LOGGER.info('-------------------------------------------')
LOGGER.info(f'Step {global_step}:')
for t in train_dataloaders.keys():
tot_ex = sum(all_gather_list(n_examples[t]))
ex_per_sec = int(tot_ex / (time() - start))
LOGGER.info(f'{t}: {tot_ex} examples trained at '
f'{ex_per_sec} ex/s')
TB_LOGGER.add_scalar(f'perf/{t}_ex_per_s', ex_per_sec,
global_step)
LOGGER.debug("after gather stats")
if global_step % opts.valid_steps == 0:
LOGGER.info('===========================================')
LOGGER.info(f"Step {global_step}: start running validation")
validate(model, val_dataloaders, opts)
LOGGER.info('===========================================')
model_saver.save(model, global_step)
# step restorer in the end to prevent missing validation checkpoint
restorer.step()
if global_step >= opts.num_train_steps:
break
LOGGER.info('===========================================')
if global_step % opts.valid_steps != 0:
LOGGER.info('===========================================')
LOGGER.info(f"Step {global_step}: start running validation")
validate(model, val_dataloaders, opts)
LOGGER.info('===========================================')
model_saver.save(model, global_step)
def _gradient_accumulation_rein(self, true_batchs, normalization):
if self.grad_accum_count > 1:
self.model.zero_grad()
for batch in true_batchs:
target_size = batch.tgt.size(0)
if self.grad_accum_count == 1:
self.model.zero_grad()
greedy_pre, sample_ids, probs, tgt = self.reinforce_batch(batch)
sample_ids = sample_ids.t().data.tolist() #batch_size*10
tgt = tgt.t().data.tolist() #batch_size*10
probs = probs.t() #batch_size*10
batch_size = probs.size(0)
rewards = []
# get the acc of every sample
for y, y_hat in zip(sample_ids, tgt): # y,y_hat len=110
rewards.append(self.get_acc(y, y_hat))
# expand the reward score to every label
rewards = torch.Tensor(rewards).unsqueeze(1).expand_as(probs)
rewards = Variable(rewards).cuda()
#greedy_pre
greedy_pre = greedy_pre.t().data.tolist()
baselines = []
for y, y_hat in zip(greedy_pre, tgt):
baselines.append(self.get_acc(y, y_hat))
baselines = torch.Tensor(baselines).unsqueeze(1).expand_as(probs)
baselines = Variable(baselines).cuda()
rewards = rewards - baselines
loss = -(probs * rewards).sum() / float(normalization)
# elif self.config.reward == 'hamming_loss':
# loss = (probs * rewards).sum() / batch_size
loss.backward()
lo = loss.item()
print('loss:' + str(lo))
# 4. Update the parameters and statistics.
if self.grad_accum_count == 1:
# Multi GPU gradient gather
if self.n_gpu > 1:
grads = [
p.grad.data for p in self.model.parameters()
if p.requires_grad and p.grad is not None
]
all_reduce_and_rescale_tensors(grads, float(1))
self.optim.step()
if self.model.decoder.state is not None:
self.model.decoder.detach_state()
# in case of multi step gradient accumulation,
# update only after accum batches
if self.grad_accum_count > 1:
if self.n_gpu > 1:
grads = [
p.grad.data for p in self.model.parameters()
if p.requires_grad and p.grad is not None
]
all_reduce_and_rescale_tensors(grads, float(1))
self.optim.step()
def main(opts):
hvd.init()
n_gpu = hvd.size()
device = torch.device("cuda", hvd.local_rank())
torch.cuda.set_device(hvd.local_rank())
rank = hvd.rank()
opts.rank = rank
LOGGER.info("device: {} n_gpu: {}, rank: {}, "
"16-bits training: {}".format(
device, n_gpu, hvd.rank(), opts.fp16))
if opts.gradient_accumulation_steps < 1:
raise ValueError("Invalid gradient_accumulation_steps parameter: {}, "
"should be >= 1".format(
opts.gradient_accumulation_steps))
set_random_seed(opts.seed)
# load DBs and image dirs
all_img_dbs = ImageLmdbGroup(opts.conf_th, opts.max_bb, opts.min_bb,
opts.num_bb, opts.compressed_db)
# train
LOGGER.info(f"Loading Train Dataset "
f"{opts.train_txt_dbs}, {opts.train_img_dbs}")
train_datasets = []
for txt_path, img_path in zip(opts.train_txt_dbs, opts.train_img_dbs):
img_db, img_db_gt = load_img_feat(img_path, all_img_dbs, opts)
qa_txt_db = VcrTxtTokLmdb(txt_path, opts.max_txt_len, task="qa")
qar_txt_db = VcrTxtTokLmdb(txt_path, opts.max_txt_len, task="qar")
train_datasets.append(
VcrDataset(qa_txt_db, img_db_gt=img_db_gt, img_db=img_db))
train_datasets.append(
VcrDataset(qar_txt_db, img_db_gt=img_db_gt, img_db=img_db))
train_dataset = ConcatDatasetWithLens(train_datasets)
train_dataloader = build_dataloader(train_dataset, vcr_collate, True, opts)
# val
LOGGER.info(f"Loading Val Dataset {opts.val_txt_db}, {opts.val_img_db}")
val_img_db, val_img_db_gt = load_img_feat(opts.val_img_db, all_img_dbs, opts)
val_txt_db = VcrTxtTokLmdb(opts.val_txt_db, -1, task="qa")
val_dataset = VcrEvalDataset(
"val", val_txt_db, img_db=val_img_db, img_db_gt=val_img_db_gt)
val_final_dataset = VcrEvalDataset(
##"test"
"val", val_txt_db, img_db=val_img_db, img_db_gt=val_img_db_gt)
val_dataloader = build_dataloader(val_dataset, vcr_eval_collate,
False, opts)
val_final_dataloader = build_dataloader(
val_final_dataset, vcr_eval_collate,
False, opts)
# Prepare model
if opts.checkpoint and opts.checkpoint_from == "pretrain":
checkpoint = torch.load(opts.checkpoint)
else:
checkpoint = {}
all_dbs = opts.train_txt_dbs + [opts.val_txt_db]
toker = json.load(open(f'{all_dbs[0]}/meta.json'))['bert']
assert all(toker == json.load(open(f'{db}/meta.json'))['bert']
for db in all_dbs)
model = UniterForVisualCommonsenseReasoning.from_pretrained(
opts.model_config, checkpoint, img_dim=IMG_DIM)
model.init_type_embedding()
model.init_type_embedding_know()
model.init_word_embedding(NUM_SPECIAL_TOKENS)
if opts.checkpoint_from == "vcr_pretrain":
checkpoint = torch.load(opts.checkpoint)
state_dict = checkpoint.get('model_state', checkpoint)
matched_state_dict = {}
unexpected_keys = set()
missing_keys = set()
for name, param in model.named_parameters():
missing_keys.add(name)
for key, data in state_dict.items():
if key in missing_keys:
matched_state_dict[key] = data
missing_keys.remove(key)
else:
unexpected_keys.add(key)
print("Unexpected_keys:", list(unexpected_keys))
print("Missing_keys:", list(missing_keys))
model.load_state_dict(matched_state_dict, strict=False)
del checkpoint
model.to(device)
# make sure every process has same model parameters in the beginning
broadcast_tensors([p.data for p in model.parameters()], 0)
set_dropout(model, opts.dropout)
# Prepare optimizer
optimizer = build_optimizer(model, opts)
model, optimizer = amp.initialize(model, optimizer,
enabled=opts.fp16, opt_level='O2')
global_step = 0
if rank == 0:
save_training_meta(opts)
TB_LOGGER.create(join(opts.output_dir, 'log'))
pbar = tqdm(total=opts.num_train_steps)
model_saver = ModelSaver(join(opts.output_dir, 'ckpt'))
os.makedirs(join(opts.output_dir, 'results')) # store VQA predictions
add_log_to_file(join(opts.output_dir, 'log', 'log.txt'))
else:
LOGGER.disabled = True
pbar = NoOp()
model_saver = NoOp()
LOGGER.info(f"***** Running training with {n_gpu} GPUs *****")
LOGGER.info(" Num examples = %d", len(train_dataset) * hvd.size())
LOGGER.info(" Batch size = %d", opts.train_batch_size)
LOGGER.info(" Accumulate steps = %d", opts.gradient_accumulation_steps)
LOGGER.info(" Num steps = %d", opts.num_train_steps)
running_loss = RunningMeter('loss')
model.train()
n_examples = 0
n_epoch = 0
start = time()
# quick hack for amp delay_unscale bug
optimizer.zero_grad()
optimizer.step()
while True:
for step, batch in enumerate(train_dataloader):
n_examples += batch['input_ids'].size(0)
loss = model(batch, compute_loss=True)
loss = loss.mean()
delay_unscale = (step+1) % opts.gradient_accumulation_steps != 0
with amp.scale_loss(loss, optimizer, delay_unscale=delay_unscale
) as scaled_loss:
scaled_loss.backward()
if not delay_unscale:
# gather gradients from every processes
# do this before unscaling to make sure every process uses
# the same gradient scale
grads = [p.grad.data for p in model.parameters()
if p.requires_grad and p.grad is not None]
all_reduce_and_rescale_tensors(grads, float(1))
running_loss(loss.item())
if (step + 1) % opts.gradient_accumulation_steps == 0:
global_step += 1
# learning rate scheduling
lr_this_step = get_lr_sched(global_step, opts)
for i, param_group in enumerate(optimizer.param_groups):
if i == 0 or i == 1:
param_group['lr'] = lr_this_step * opts.lr_mul
elif i == 2 or i == 3:
param_group['lr'] = lr_this_step
else:
raise ValueError()
TB_LOGGER.add_scalar('lr', lr_this_step, global_step)
# log loss
# NOTE: not gathered across GPUs for efficiency
TB_LOGGER.add_scalar('loss', running_loss.val, global_step)
TB_LOGGER.step()
# update model params
if opts.grad_norm != -1:
grad_norm = clip_grad_norm_(amp.master_params(optimizer),
opts.grad_norm)
TB_LOGGER.add_scalar('grad_norm', grad_norm, global_step)
optimizer.step()
optimizer.zero_grad()
pbar.update(1)
if global_step % 100 == 0:
# monitor training throughput
LOGGER.info(f'============Step {global_step}=============')
tot_ex = sum(all_gather_list(n_examples))
ex_per_sec = int(tot_ex / (time()-start))
LOGGER.info(f'{tot_ex} examples trained at '
f'{ex_per_sec} ex/s')
TB_LOGGER.add_scalar('perf/ex_per_s',
ex_per_sec, global_step)
LOGGER.info(f'===========================================')
if global_step % opts.valid_steps == 0:
val_log, results = validate(
model, val_dataloader)
TB_LOGGER.log_scaler_dict(val_log)
model_saver.save(model, global_step)
if global_step >= opts.num_train_steps:
break
if global_step >= opts.num_train_steps:
break
n_epoch += 1
LOGGER.info(f"finished {n_epoch} epochs")
if global_step % opts.valid_steps != 0:
val_log, results = validate(
model, val_dataloader)
TB_LOGGER.log_scaler_dict(val_log)
val_log, results = validate(model, val_final_dataloader)
with open(f'{opts.output_dir}/results/'
f'results_{global_step}_final_qa_qar_'
f'rank{rank}.json', 'w') as f:
json.dump(results, f)
TB_LOGGER.log_scaler_dict(val_log)
model_saver.save(model, global_step)
def main(opts):
hvd.init()
n_gpu = hvd.size()
device = torch.device("cuda", hvd.local_rank())
torch.cuda.set_device(hvd.local_rank())
rank = hvd.rank()
opts.rank = rank
LOGGER.info("device: {} n_gpu: {}, rank: {}, "
"16-bits training: {}".format(device, n_gpu, hvd.rank(),
opts.fp16))
if opts.gradient_accumulation_steps < 1:
raise ValueError("Invalid gradient_accumulation_steps parameter: {}, "
"should be >= 1".format(
opts.gradient_accumulation_steps))
set_random_seed(opts.seed)
# train_examples = None
LOGGER.info(f"Loading Train Dataset {opts.train_txt_db}, "
f"{opts.train_img_db}")
train_dataloader = create_dataloader(
opts.train_img_db,
opts.train_txt_db,
opts.train_batch_size,
True,
VeDataset,
ve_collate,
opts,
)
val_dataloader = create_dataloader(
opts.val_img_db,
opts.val_txt_db,
opts.val_batch_size,
False,
VeEvalDataset,
ve_eval_collate,
opts,
)
test_dataloader = create_dataloader(
opts.test_img_db,
opts.test_txt_db,
opts.val_batch_size,
False,
VeEvalDataset,
ve_eval_collate,
opts,
)
# Prepare model
if opts.checkpoint:
checkpoint = torch.load(opts.checkpoint)
else:
checkpoint = {}
bert_model = json.load(open(f"{opts.train_txt_db}/meta.json"))["bert"]
if "bert" not in bert_model:
bert_model = "bert-large-cased" # quick hack for glove exp
model = UniterForVisualEntailment.from_pretrained(opts.model_config,
state_dict=checkpoint,
img_dim=IMG_DIM)
model.to(device)
# make sure every process has same model parameters in the beginning
broadcast_tensors([p.data for p in model.parameters()], 0)
set_dropout(model, opts.dropout)
# Prepare optimizer
optimizer = build_optimizer(model, opts)
model, optimizer = amp.initialize(model,
optimizer,
enabled=opts.fp16,
opt_level="O2")
global_step = 0
if rank == 0:
save_training_meta(opts)
TB_LOGGER.create(join(opts.output_dir, "log"))
pbar = tqdm(total=opts.num_train_steps)
model_saver = ModelSaver(join(opts.output_dir, "ckpt"))
pickle.dump(ans2label,
open(join(opts.output_dir, "ckpt", "ans2label.pkl"), "wb"))
os.makedirs(join(opts.output_dir, "results")) # store VQA predictions
add_log_to_file(join(opts.output_dir, "log", "log.txt"))
else:
LOGGER.disabled = True
pbar = NoOp()
model_saver = NoOp()
LOGGER.info(f"***** Running training with {n_gpu} GPUs *****")
LOGGER.info(" Num examples = %d", len(train_dataloader.dataset))
LOGGER.info(" Batch size = %d", opts.train_batch_size)
LOGGER.info(" Accumulate steps = %d", opts.gradient_accumulation_steps)
LOGGER.info(" Num steps = %d", opts.num_train_steps)
running_loss = RunningMeter("loss")
model.train()
n_examples = 0
n_epoch = 0
start = time()
# quick hack for amp delay_unscale bug
optimizer.zero_grad()
optimizer.step()
while True:
for step, batch in enumerate(train_dataloader):
n_examples += batch["input_ids"].size(0)
loss = model(batch, compute_loss=True)
loss = loss.mean() * batch["targets"].size(1) # instance-leval bce
delay_unscale = (step + 1) % opts.gradient_accumulation_steps != 0
with amp.scale_loss(loss, optimizer,
delay_unscale=delay_unscale) as scaled_loss:
scaled_loss.backward()
if not delay_unscale:
# gather gradients from every processes
# do this before unscaling to make sure every process uses
# the same gradient scale
grads = [
p.grad.data for p in model.parameters()
if p.requires_grad and p.grad is not None
]
all_reduce_and_rescale_tensors(grads, float(1))
running_loss(loss.item())
if (step + 1) % opts.gradient_accumulation_steps == 0:
global_step += 1
# learning rate scheduling
lr_this_step = get_lr_sched(global_step, opts)
for param_group in optimizer.param_groups:
param_group["lr"] = lr_this_step
TB_LOGGER.add_scalar("lr", lr_this_step, global_step)
# log loss
# NOTE: not gathered across GPUs for efficiency
TB_LOGGER.add_scalar("loss", running_loss.val, global_step)
TB_LOGGER.step()
# update model params
if opts.grad_norm != -1:
grad_norm = clip_grad_norm_(amp.master_params(optimizer),
opts.grad_norm)
TB_LOGGER.add_scalar("grad_norm", grad_norm, global_step)
optimizer.step()
optimizer.zero_grad()
pbar.update(1)
if global_step % 100 == 0:
# monitor training throughput
LOGGER.info(f"============Step {global_step}=============")
tot_ex = sum(all_gather_list(n_examples))
ex_per_sec = int(tot_ex / (time() - start))
LOGGER.info(f"{tot_ex} examples trained at "
f"{ex_per_sec} ex/s")
TB_LOGGER.add_scalar("perf/ex_per_s", ex_per_sec,
global_step)
LOGGER.info(f"===========================================")
if global_step % opts.valid_steps == 0:
for split, loader in [
("val", val_dataloader),
("test", test_dataloader),
]:
LOGGER.info(f"Step {global_step}: start running "
f"validation on {split} split...")
val_log, results = validate(model, loader, label2ans,
split)
with open(
f"{opts.output_dir}/results/"
f"{split}_results_{global_step}_"
f"rank{rank}.json",
"w",
) as f:
json.dump(results, f)
TB_LOGGER.log_scaler_dict(val_log)
model_saver.save(model, global_step)
if global_step >= opts.num_train_steps:
break
if global_step >= opts.num_train_steps:
break
n_epoch += 1
LOGGER.info(f"Step {global_step}: finished {n_epoch} epochs")
if opts.num_train_steps % opts.valid_steps != 0:
for split, loader in [("val", val_dataloader),
("test", test_dataloader)]:
LOGGER.info(f"Step {global_step}: start running "
f"validation on {split} split...")
val_log, results = validate(model, loader, label2ans, split)
with open(
f"{opts.output_dir}/results/"
f"{split}_results_{global_step}_"
f"rank{rank}_final.json",
"w",
) as f:
json.dump(results, f)
TB_LOGGER.log_scaler_dict(val_log)
model_saver.save(model, global_step)
def main(opts):
hvd.init()
n_gpu = hvd.size()
device = torch.device("cuda", hvd.local_rank())
torch.cuda.set_device(hvd.local_rank())
rank = hvd.rank()
opts.rank = rank
LOGGER.info("device: {} n_gpu: {}, rank: {}, "
"16-bits training: {}".format(device, n_gpu, hvd.rank(),
opts.fp16))
if opts.gradient_accumulation_steps < 1:
raise ValueError("Invalid gradient_accumulation_steps parameter: {}, "
"should be >= 1".format(
opts.gradient_accumulation_steps))
set_random_seed(opts.seed)
if hasattr(opts, 'ans2label_path'):
ans2label = json.load(open(opts.ans2label_path, 'r', encoding='utf-8'))
else:
ans2label = json.load(
open(f'{dirname(abspath(__file__))}'
f'/utils/ans2label.json'))
label2ans = {label: ans for ans, label in ans2label.items()}
# load DBs and image dirs
all_img_dbs = ImageLmdbGroup(opts.conf_th, opts.max_bb, opts.min_bb,
opts.num_bb, opts.compressed_db)
# train
LOGGER.info(f"Loading Train Dataset "
f"{opts.train_txt_dbs}, {opts.train_img_dbs}")
train_datasets = []
for txt_path, img_path in zip(opts.train_txt_dbs, opts.train_img_dbs):
img_db = all_img_dbs[img_path]
txt_db = TxtTokLmdb(txt_path, opts.max_txt_len)
train_datasets.append(VqaDataset(len(ans2label), txt_db, img_db))
train_dataset = ConcatDatasetWithLens(train_datasets)
train_dataloader = build_dataloader(train_dataset, vqa_collate, True, opts)
# val
LOGGER.info(f"Loading Train Dataset {opts.val_txt_db}, {opts.val_img_db}")
val_img_db = all_img_dbs[opts.val_img_db]
val_txt_db = TxtTokLmdb(opts.val_txt_db, -1)
val_dataset = VqaEvalDataset(len(ans2label), val_txt_db, val_img_db)
val_dataloader = build_dataloader(val_dataset, vqa_eval_collate, False,
opts)
# Prepare model
if opts.checkpoint:
checkpoint = torch.load(opts.checkpoint, map_location=device)
else:
checkpoint = {}
all_dbs = opts.train_txt_dbs + [opts.val_txt_db]
toker = json.load(open(f'{all_dbs[0]}/meta.json'))['bert']
assert all(toker == json.load(open(f'{db}/meta.json'))['bert']
for db in all_dbs)
model = UniterForVisualQuestionAnswering.from_pretrained(
opts.model_config,
checkpoint,
img_dim=IMG_DIM,
num_answer=len(ans2label))
model.to(device)
# make sure every process has same model parameters in the beginning
broadcast_tensors([p.data for p in model.parameters()], 0)
set_dropout(model, opts.dropout)
# Prepare optimizer
optimizer = build_optimizer(model, opts)
model, optimizer = amp.initialize(model,
optimizer,
enabled=opts.fp16,
opt_level='O2')
global_step = 0
if rank == 0:
save_training_meta(opts)
TB_LOGGER.create(join(opts.output_dir, 'log'))
pbar = tqdm(total=opts.num_train_steps)
model_saver = ModelSaver(join(opts.output_dir, 'ckpt'))
json.dump(ans2label,
open(join(opts.output_dir, 'ckpt', 'ans2label.json'), 'w'))
os.makedirs(join(opts.output_dir, 'results')) # store VQA predictions
add_log_to_file(join(opts.output_dir, 'log', 'log.txt'))
else:
LOGGER.disabled = True
pbar = NoOp()
model_saver = NoOp()
LOGGER.info(f"***** Running training with {n_gpu} GPUs *****")
LOGGER.info(" Num examples = %d", len(train_dataset) * hvd.size())
LOGGER.info(" Batch size = %d", opts.train_batch_size)
LOGGER.info(" Accumulate steps = %d", opts.gradient_accumulation_steps)
LOGGER.info(" Num steps = %d", opts.num_train_steps)
running_loss = RunningMeter('loss')
model.train()
n_examples = 0
n_epoch = 0
start = time()
# quick hack for amp delay_unscale bug
optimizer.zero_grad()
optimizer.step()
while True:
for step, batch in enumerate(train_dataloader):
n_examples += batch['input_ids'].size(0)
loss = model(batch, compute_loss=True)
loss = loss.mean() * batch['targets'].size(1) # instance-leval bce
delay_unscale = (step + 1) % opts.gradient_accumulation_steps != 0
with amp.scale_loss(loss, optimizer,
delay_unscale=delay_unscale) as scaled_loss:
scaled_loss.backward()
if not delay_unscale:
# gather gradients from every processes
# do this before unscaling to make sure every process uses
# the same gradient scale
grads = [
p.grad.data for p in model.parameters()
if p.requires_grad and p.grad is not None
]
all_reduce_and_rescale_tensors(grads, float(1))
running_loss(loss.item())
if (step + 1) % opts.gradient_accumulation_steps == 0:
global_step += 1
# learning rate scheduling
lr_this_step = get_lr_sched(global_step, opts)
for i, param_group in enumerate(optimizer.param_groups):
if i == 0 or i == 1:
param_group['lr'] = lr_this_step * opts.lr_mul
elif i == 2 or i == 3:
param_group['lr'] = lr_this_step
else:
raise ValueError()
TB_LOGGER.add_scalar('lr', lr_this_step, global_step)
# log loss
# NOTE: not gathered across GPUs for efficiency
TB_LOGGER.add_scalar('loss', running_loss.val, global_step)
TB_LOGGER.step()
# update model params
if opts.grad_norm != -1:
grad_norm = clip_grad_norm_(amp.master_params(optimizer),
opts.grad_norm)
TB_LOGGER.add_scalar('grad_norm', grad_norm, global_step)
optimizer.step()
optimizer.zero_grad()
pbar.update(1)
if global_step % 100 == 0:
# monitor training throughput
LOGGER.info(f'============Step {global_step}=============')
tot_ex = sum(all_gather_list(n_examples))
ex_per_sec = int(tot_ex / (time() - start))
LOGGER.info(f'{tot_ex} examples trained at '
f'{ex_per_sec} ex/s')
TB_LOGGER.add_scalar('perf/ex_per_s', ex_per_sec,
global_step)
LOGGER.info(f'===========================================')
if global_step % opts.valid_steps == 0:
val_log, results = validate(model, val_dataloader,
label2ans)
with open(
f'{opts.output_dir}/results/'
f'results_{global_step}_'
f'rank{rank}.json', 'w') as f:
json.dump(results, f)
TB_LOGGER.log_scaler_dict(val_log)
model_saver.save(model, global_step)
if global_step >= opts.num_train_steps:
break
if global_step >= opts.num_train_steps:
break
n_epoch += 1
LOGGER.info(f"finished {n_epoch} epochs")
if opts.num_train_steps % opts.valid_steps != 0:
val_log, results = validate(model, val_dataloader, label2ans)
with open(
f'{opts.output_dir}/results/'
f'results_{global_step}_'
f'rank{rank}.json', 'w') as f:
json.dump(results, f)
TB_LOGGER.log_scaler_dict(val_log)
model_saver.save(model, global_step)
def _gradient_accumulation(self, true_batchs, normalization, total_stats,
report_stats):
if self.grad_accum_count > 1:
self.model.zero_grad()
for batch in true_batchs:
target_size = batch.tgt.size(0)
# Truncated BPTT: reminder not compatible with accum > 1
if self.trunc_size:
trunc_size = self.trunc_size
else:
trunc_size = target_size
# dec_state = None
src = make_features(batch, 'src') # src 12 * 146 维度
_, src_lengths = batch.src
tgt_outer = make_features(batch, 'tgt')
structure1 = make_features(batch, 'structure1')
structure1 = structure1.transpose(0, 1)
structure1 = structure1.transpose(1, 2)
structure2 = make_features(batch, 'structure2')
structure2 = structure2.transpose(0, 1)
structure2 = structure2.transpose(1, 2)
structure3 = make_features(batch, 'structure3')
structure3 = structure3.transpose(0, 1)
structure3 = structure3.transpose(1, 2)
structure4 = make_features(batch, 'structure4')
structure4 = structure4.transpose(0, 1)
structure4 = structure4.transpose(1, 2)
structure5 = make_features(batch, 'structure5')
structure5 = structure5.transpose(0, 1)
structure5 = structure5.transpose(1, 2)
# structure6 = make_features(batch, 'structure6')
# structure6 = structure6.transpose(0, 1)
# structure6 = structure6.transpose(1, 2)
#
# structure7 = make_features(batch, 'structure7')
# structure7 = structure7.transpose(0, 1)
# structure7 = structure7.transpose(1, 2)
#
# structure8 = make_features(batch, 'structure8')
# structure8 = structure8.transpose(0, 1)
# structure8 = structure8.transpose(1, 2)
for j in range(0, target_size - 1, trunc_size):
# 1. Create truncated target.
tgt = tgt_outer[j:j + trunc_size]
# 2. F-prop all but generator.
if self.grad_accum_count == 1:
self.model.zero_grad()
outputs, attns = self.model(src, tgt, structure1, structure2,
structure3, structure4, structure5,
src_lengths)
# 3. Compute loss in shards for memory efficiency.
batch_stats = self.train_loss.sharded_compute_loss(
batch, outputs, attns, j, trunc_size, self.shard_size,
normalization)
total_stats.update(batch_stats)
report_stats.update(batch_stats)
# 4. Update the parameters and statistics.
if self.grad_accum_count == 1:
# Multi GPU gradient gather
if self.n_gpu > 1:
grads = [
p.grad.data for p in self.model.parameters()
if p.requires_grad and p.grad is not None
]
all_reduce_and_rescale_tensors(grads, float(1))
self.optim.step()
# If truncated, don't backprop fully.
# TO CHECK
# if dec_state is not None:
# dec_state.detach()
if self.model.decoder.state is not None:
self.model.decoder.detach_state()
# in case of multi step gradient accumulation,
# update only after accum batches
if self.grad_accum_count > 1:
if self.n_gpu > 1:
grads = [
p.grad.data for p in self.model.parameters()
if p.requires_grad and p.grad is not None
]
all_reduce_and_rescale_tensors(grads, float(1))
self.optim.step()
def main(opts):
hvd.init()
n_gpu = hvd.size()
device = torch.device("cuda", hvd.local_rank())
torch.cuda.set_device(hvd.local_rank())
rank = hvd.rank()
opts.rank = rank
LOGGER.info("device: {} n_gpu: {}, rank: {}, "
"16-bits training: {}".format(
device, n_gpu, hvd.rank(), opts.fp16))
if opts.gradient_accumulation_steps < 1:
raise ValueError("Invalid gradient_accumulation_steps parameter: {}, "
"should be >= 1".format(
opts.gradient_accumulation_steps))
set_random_seed(opts.seed)
if rank == 0:
save_training_meta(opts)
TB_LOGGER.create(join(opts.output_dir, 'log'))
pbar = tqdm(total=opts.num_train_steps)
model_saver = ModelSaver(join(opts.output_dir, 'ckpt'))
add_log_to_file(join(opts.output_dir, 'log', 'log.txt'))
else:
LOGGER.disabled = True
pbar = NoOp()
model_saver = NoOp()
all_dbs = [db for datasets in [opts.train_datasets, opts.val_datasets]
for dset in datasets for db in dset['db']]
tokenizer = json.load(open(f'{all_dbs[0]}/meta.json'))['bert']
#print(tokenizer)
# assert all(tokenizer == json.load(open(f'{db}/meta.json'))['bert']
# for db in all_dbs)
# build data loaders
train_dataloaders, all_img_dbs = create_dataloaders(
opts.train_datasets, True, opts)
val_dataloaders, _ = create_dataloaders(
opts.val_datasets, False, opts, all_img_dbs)
meta_loader = MetaLoader(train_dataloaders,
accum_steps=opts.gradient_accumulation_steps,
distributed=n_gpu > 1)
meta_loader = PrefetchLoader(meta_loader)
# Prepare model
if opts.checkpoint:
checkpoint = torch.load(opts.checkpoint)
else:
checkpoint = {}
if opts.rename_checkpoints:
rename_checkpoint(checkpoint)
#Include early_adaptation
if opts.early_adaptation:
early_adaptation_checkpoint = torch.load(opts.early_adaptation_checkpoint)
checkpoint['roberta.img_embeddings.img_linear.weight'] = early_adaptation_checkpoint['v2w_linear.weight']
checkpoint['roberta.img_embeddings.img_linear.bias'] = early_adaptation_checkpoint['v2w_linear.bias']
model = VLXLMRForPretraining.from_pretrained(
opts.model_config, checkpoint,
img_dim=IMG_DIM, img_label_dim=IMG_LABEL_DIM,
nce_temp=opts.nce_temp, ot_pos_only=opts.ot_pos_only)
# model = UniterForPretraining.from_pretrained(
# opts.model_config, checkpoint,
# img_dim=IMG_DIM, img_label_dim=IMG_LABEL_DIM,
# nce_temp=opts.nce_temp, ot_pos_only=opts.ot_pos_only)
model.pad_vocab() # tensor core padding for vocabulary
model.to(device)
model.train()
# make sure every process has same model parameters in the beginning
broadcast_tensors([p.data for p in model.parameters()], 0)
set_dropout(model, opts.dropout)
# Prepare optimizer
optimizer = build_optimizer(model, opts)
task2scaler = {t: i for i, t in enumerate(train_dataloaders.keys())}
model, optimizer = amp.initialize(model, optimizer,
num_losses=len(task2scaler),
enabled=opts.fp16, opt_level='O2')
#global_step = 0
#Initialize the TrainingRestorer
restorer = TrainingRestorer(opts, model, optimizer)
global_step = restorer.global_step
TB_LOGGER._global_step = global_step
if hvd.rank() !=0:
restorer = NoOp() #Added for Restoring the Checkpoints
if global_step > 0:
pbar.update(global_step)
LOGGER.info(f"***** Running training with {n_gpu} GPUs *****")
LOGGER.info(" Batch size = %d", opts.train_batch_size)
LOGGER.info(" Accumulate steps = %d", opts.gradient_accumulation_steps)
LOGGER.info(" Num steps = %d", opts.num_train_steps)
# to compute training statistics
task2loss = {task: RunningMeter(f'loss/{task}')
for task in train_dataloaders.keys()}
# ITM w/ OT
if opts.itm_ot_lambda > 0:
for task in train_dataloaders.keys():
if task.startswith('itm'):
task2loss[f'{task}_xe'] = RunningMeter(f'loss/{task}_xe')
task2loss[f'{task}_ot'] = RunningMeter(f'loss/{task}_ot')
if not opts.ot_pos_only:
task2loss[f'{task}_ot_pos'] = RunningMeter(
f'loss/{task}_ot_pos')
task2loss[f'{task}_ot_neg'] = RunningMeter(
f'loss/{task}_ot_neg')
n_examples = defaultdict(int)
n_in_units = defaultdict(int)
n_loss_units = defaultdict(int)
n_neg_nce = defaultdict(int)
grad_norm = 0
start = time()
#Added by Mingyang to debug the training procedure
# debug_start = torch.cuda.Event(enable_timing=True)
# debug_end = torch.cuda.Event(enable_timing=True)
# quick hack for amp delay_unscale bug
optimizer.zero_grad()
optimizer.step()
#Added by Mingyang Zhou
# debug_start.record()
for step, (name, batch) in enumerate(meta_loader):
# forward pass
assert all(name == n for n in all_gather_list(name))
n_examples[name] += batch['input_ids'].size(0)
n_in_units[name] += (batch['attn_masks'] == 1).sum().item()
if 'nce' in name:
n_neg_nce[name] += batch['neg_feats'].size(0)
task = name.split('_')[0]
loss = model(batch, task=task, compute_loss=True)
if task.startswith('itm'):
# OT
itm_loss, ot_loss = loss
n_loss_units[name] += itm_loss.size(0)
itm_loss = itm_loss.mean()
if ot_loss is not None:
if not opts.ot_pos_only:
ot_pos, ot_neg = ot_loss
ot_loss = (ot_pos.sum() - ot_neg.sum()
) / (ot_pos.size(0) + ot_neg.size(0))
# NOTE: be ware of empty tensor
ot_pos = ot_pos.mean().item()
if not math.isnan(ot_pos):
task2loss[f'{name}_ot_pos'](ot_pos)
ot_neg = ot_neg.mean().item()
if not math.isnan(ot_neg):
task2loss[f'{name}_ot_neg'](ot_neg)
else:
ot_loss = ot_loss.mean()
loss = itm_loss + opts.itm_ot_lambda * ot_loss
task2loss[f'{name}_xe'](itm_loss.item())
task2loss[f'{name}_ot'](ot_loss.item())
else:
loss = itm_loss
elif task.startswith('vmlm-soft'):
loss = 1000*loss.mean()
else:
n_loss_units[name] += loss.size(0)
loss = loss.mean() # loss is not normalized in model
# backward pass
delay_unscale = (step+1) % opts.gradient_accumulation_steps != 0
with amp.scale_loss(loss, optimizer, delay_unscale=delay_unscale,
loss_id=task2scaler[name]) as scaled_loss:
scaled_loss.backward()
if not delay_unscale:
# gather gradients from every processes
# do this before unscaling to make sure every process uses
# the same gradient scale
grads = [p.grad.data for p in model.parameters()
if p.requires_grad and p.grad is not None]
all_reduce_and_rescale_tensors(grads, float(1))
task2loss[name](loss.item())
# optimizer update and logging
if (step + 1) % opts.gradient_accumulation_steps == 0:
global_step += 1
# learning rate scheduling
lr_this_step = get_lr_sched(global_step, opts)
for param_group in optimizer.param_groups:
param_group['lr'] = lr_this_step
TB_LOGGER.add_scalar('lr', lr_this_step, global_step)
# log loss
# for t, l in task2loss.items():
# loss = sum(v for v in all_gather_list(l.val)
# if v is not None) / hvd.size()
# task2loss[t] = RunningMeter(f'loss/{t}', loss)
TB_LOGGER.log_scaler_dict({l.name: l.val
for l in task2loss.values()
if l.val is not None})
TB_LOGGER.step()
# update model params
if opts.grad_norm != -1:
'''
if global_step % 10 == 0 and not opts.fp16:
bias = model.bert.img_embeddings.img_linear.bias
weight = model.bert.img_embeddings.img_linear.weight
print(f"bnorm: {bias.norm()}")
print(f"wnorm: {weight.norm()}")
print(f"bgnorm: {bias.grad.norm()}")
print(f"wgnorm: {weight.grad.norm()}")
mask = model.bert.img_embeddings.mask_embedding.weight
print(f"mnorm: {mask.norm()}")
print(f"mgnorm: {mask.grad.norm()}")
print([(n, p.grad.norm().item())
for n, p in model.named_parameters()
if p.grad is not None
and p.grad.norm().item() > grad_norm/10])
'''
grad_norm = clip_grad_norm_(amp.master_params(optimizer),
opts.grad_norm)
TB_LOGGER.add_scalar('grad_norm', grad_norm, global_step)
optimizer.step()
optimizer.zero_grad()
pbar.update(1)
if global_step % 100 == 0:
# monitor training throughput
LOGGER.info(f'==============Step {global_step}===============')
for t in train_dataloaders.keys():
assert all(tt == t for tt in all_gather_list(t))
tot_ex = sum(all_gather_list(n_examples[t]))
ex_per_sec = int(tot_ex / (time()-start))
tot_in = sum(all_gather_list(n_in_units[t]))
in_per_sec = int(tot_in / (time()-start))
tot_l = sum(all_gather_list(n_loss_units[t]))
l_per_sec = int(tot_l / (time()-start))
LOGGER.info(f'{t}: {tot_ex} examples trained at '
f'{ex_per_sec} ex/s')
TB_LOGGER.add_scalar(f'perf/{t}_ex_per_s', ex_per_sec,
global_step)
TB_LOGGER.add_scalar(f'perf/{t}_in_per_s', in_per_sec,
global_step)
TB_LOGGER.add_scalar(f'perf/{t}_loss_per_s', l_per_sec,
global_step)
if 'nce' in t:
avg_neg = sum(all_gather_list(n_neg_nce[t])
) / hvd.size() // step
LOGGER.info(f'{t}: averaging '
f'{avg_neg} negative samples')
LOGGER.info(f'===============================================')
if global_step % opts.valid_steps == 0:
LOGGER.info(f'Step {global_step}: start validation')
validate(model, val_dataloaders)
#os.makedir('/'.join([opts.output_dir, "ckpt")
model_saver.save(model, global_step, optimizer)
restorer.step()
if global_step >= opts.num_train_steps:
break
if global_step % opts.valid_steps != 0:
LOGGER.info(f'Step {global_step}: start validation')
validate(model, val_dataloaders)
model_saver.save(model, global_step)
