382 lines
12 KiB
Python
382 lines
12 KiB
Python
import io
|
|
import os
|
|
import math
|
|
import time
|
|
import json
|
|
import glob
|
|
from collections import defaultdict, deque, OrderedDict
|
|
import datetime
|
|
import numpy as np
|
|
|
|
|
|
from pathlib import Path
|
|
import argparse
|
|
|
|
import torch
|
|
from torch import optim as optim
|
|
import torch.distributed as dist
|
|
#from tensorboardX import SummaryWriter
|
|
|
|
|
|
def is_dist_avail_and_initialized():
|
|
if not dist.is_available():
|
|
return False
|
|
if not dist.is_initialized():
|
|
return False
|
|
return True
|
|
|
|
|
|
def get_world_size():
|
|
if not is_dist_avail_and_initialized():
|
|
return 1
|
|
return dist.get_world_size()
|
|
|
|
|
|
def get_rank():
|
|
if not is_dist_avail_and_initialized():
|
|
return 0
|
|
return dist.get_rank()
|
|
|
|
|
|
def is_main_process():
|
|
return get_rank() == 0
|
|
|
|
|
|
def save_on_master(*args, **kwargs):
|
|
if is_main_process():
|
|
torch.save(*args, **kwargs)
|
|
|
|
|
|
def setup_for_distributed(is_master):
|
|
"""
|
|
This function disables printing when not in master process
|
|
"""
|
|
import builtins as __builtin__
|
|
builtin_print = __builtin__.print
|
|
|
|
def print(*args, **kwargs):
|
|
force = kwargs.pop('force', False)
|
|
if is_master or force:
|
|
builtin_print(*args, **kwargs)
|
|
|
|
__builtin__.print = print
|
|
|
|
|
|
def init_distributed_mode(args):
|
|
if int(os.getenv('OMPI_COMM_WORLD_SIZE', '0')) > 0:
|
|
rank = int(os.environ['OMPI_COMM_WORLD_RANK'])
|
|
local_rank = int(os.environ['OMPI_COMM_WORLD_LOCAL_RANK'])
|
|
world_size = int(os.environ['OMPI_COMM_WORLD_SIZE'])
|
|
|
|
os.environ["LOCAL_RANK"] = os.environ['OMPI_COMM_WORLD_LOCAL_RANK']
|
|
os.environ["RANK"] = os.environ['OMPI_COMM_WORLD_RANK']
|
|
os.environ["WORLD_SIZE"] = os.environ['OMPI_COMM_WORLD_SIZE']
|
|
|
|
args.rank = int(os.environ["RANK"])
|
|
args.world_size = int(os.environ["WORLD_SIZE"])
|
|
args.gpu = int(os.environ["LOCAL_RANK"])
|
|
|
|
elif 'RANK' in os.environ and 'WORLD_SIZE' in os.environ:
|
|
args.rank = int(os.environ["RANK"])
|
|
args.world_size = int(os.environ['WORLD_SIZE'])
|
|
args.gpu = int(os.environ['LOCAL_RANK'])
|
|
|
|
else:
|
|
print('Not using distributed mode')
|
|
args.distributed = False
|
|
return
|
|
|
|
args.distributed = True
|
|
args.dist_backend = 'nccl'
|
|
args.dist_url = "env://"
|
|
print('| distributed init (rank {}): {}, gpu {}'.format(
|
|
args.rank, args.dist_url, args.gpu), flush=True)
|
|
|
|
|
|
def cosine_scheduler(base_value, final_value, epochs, niter_per_ep, warmup_epochs=0,
|
|
start_warmup_value=0, warmup_steps=-1):
|
|
warmup_schedule = np.array([])
|
|
warmup_iters = warmup_epochs * niter_per_ep
|
|
if warmup_steps > 0:
|
|
warmup_iters = warmup_steps
|
|
print("Set warmup steps = %d" % warmup_iters)
|
|
if warmup_epochs > 0:
|
|
warmup_schedule = np.linspace(start_warmup_value, base_value, warmup_iters)
|
|
|
|
iters = np.arange(epochs * niter_per_ep - warmup_iters)
|
|
schedule = np.array(
|
|
[final_value + 0.5 * (base_value - final_value) * (1 + math.cos(math.pi * i / (len(iters)))) for i in iters])
|
|
|
|
schedule = np.concatenate((warmup_schedule, schedule))
|
|
|
|
assert len(schedule) == epochs * niter_per_ep
|
|
return schedule
|
|
|
|
|
|
def constant_scheduler(base_value, epochs, niter_per_ep, warmup_epochs=0,
|
|
start_warmup_value=1e-6, warmup_steps=-1):
|
|
warmup_schedule = np.array([])
|
|
warmup_iters = warmup_epochs * niter_per_ep
|
|
if warmup_steps > 0:
|
|
warmup_iters = warmup_steps
|
|
print("Set warmup steps = %d" % warmup_iters)
|
|
if warmup_iters > 0:
|
|
warmup_schedule = np.linspace(start_warmup_value, base_value, warmup_iters)
|
|
|
|
iters = epochs * niter_per_ep - warmup_iters
|
|
schedule = np.array([base_value] * iters)
|
|
|
|
schedule = np.concatenate((warmup_schedule, schedule))
|
|
|
|
assert len(schedule) == epochs * niter_per_ep
|
|
return schedule
|
|
|
|
|
|
def get_parameter_groups(model, weight_decay=1e-5, base_lr=1e-4, skip_list=(), get_num_layer=None, get_layer_scale=None, **kwargs):
|
|
parameter_group_names = {}
|
|
parameter_group_vars = {}
|
|
|
|
for name, param in model.named_parameters():
|
|
if not param.requires_grad:
|
|
continue # frozen weights
|
|
if len(kwargs.get('filter_name', [])) > 0:
|
|
flag = False
|
|
for filter_n in kwargs.get('filter_name', []):
|
|
if filter_n in name:
|
|
print(f"filter {name} because of the pattern {filter_n}")
|
|
flag = True
|
|
if flag:
|
|
continue
|
|
|
|
default_scale=1.
|
|
|
|
if param.ndim <= 1 or name.endswith(".bias") or name in skip_list: # param.ndim <= 1 len(param.shape) == 1
|
|
group_name = "no_decay"
|
|
this_weight_decay = 0.
|
|
else:
|
|
group_name = "decay"
|
|
this_weight_decay = weight_decay
|
|
|
|
if get_num_layer is not None:
|
|
layer_id = get_num_layer(name)
|
|
group_name = "layer_%d_%s" % (layer_id, group_name)
|
|
else:
|
|
layer_id = None
|
|
|
|
if group_name not in parameter_group_names:
|
|
if get_layer_scale is not None:
|
|
scale = get_layer_scale(layer_id)
|
|
else:
|
|
scale = default_scale
|
|
|
|
parameter_group_names[group_name] = {
|
|
"weight_decay": this_weight_decay,
|
|
"params": [],
|
|
"lr": base_lr,
|
|
"lr_scale": scale,
|
|
}
|
|
|
|
parameter_group_vars[group_name] = {
|
|
"weight_decay": this_weight_decay,
|
|
"params": [],
|
|
"lr": base_lr,
|
|
"lr_scale": scale,
|
|
}
|
|
|
|
parameter_group_vars[group_name]["params"].append(param)
|
|
parameter_group_names[group_name]["params"].append(name)
|
|
|
|
print("Param groups = %s" % json.dumps(parameter_group_names, indent=2))
|
|
return list(parameter_group_vars.values())
|
|
|
|
|
|
def create_optimizer(args, model, get_num_layer=None, get_layer_scale=None, filter_bias_and_bn=True, skip_list=None, **kwargs):
|
|
opt_lower = args.opt.lower()
|
|
weight_decay = args.weight_decay
|
|
|
|
skip = {}
|
|
if skip_list is not None:
|
|
skip = skip_list
|
|
elif hasattr(model, 'no_weight_decay'):
|
|
skip = model.no_weight_decay()
|
|
print(f"Skip weight decay name marked in model: {skip}")
|
|
parameters = get_parameter_groups(model, weight_decay, args.lr, skip, get_num_layer, get_layer_scale, **kwargs)
|
|
weight_decay = 0.
|
|
|
|
if 'fused' in opt_lower:
|
|
assert has_apex and torch.cuda.is_available(), 'APEX and CUDA required for fused optimizers'
|
|
|
|
opt_args = dict(lr=args.lr, weight_decay=weight_decay)
|
|
if hasattr(args, 'opt_eps') and args.opt_eps is not None:
|
|
opt_args['eps'] = args.opt_eps
|
|
if hasattr(args, 'opt_beta1') and args.opt_beta1 is not None:
|
|
opt_args['betas'] = (args.opt_beta1, args.opt_beta2)
|
|
|
|
print('Optimizer config:', opt_args)
|
|
opt_split = opt_lower.split('_')
|
|
opt_lower = opt_split[-1]
|
|
if opt_lower == 'sgd' or opt_lower == 'nesterov':
|
|
opt_args.pop('eps', None)
|
|
optimizer = optim.SGD(parameters, momentum=args.momentum, nesterov=True, **opt_args)
|
|
elif opt_lower == 'momentum':
|
|
opt_args.pop('eps', None)
|
|
optimizer = optim.SGD(parameters, momentum=args.momentum, nesterov=False, **opt_args)
|
|
elif opt_lower == 'adam':
|
|
optimizer = optim.Adam(parameters, **opt_args)
|
|
elif opt_lower == 'adamw':
|
|
optimizer = optim.AdamW(parameters, **opt_args)
|
|
elif opt_lower == 'adadelta':
|
|
optimizer = optim.Adadelta(parameters, **opt_args)
|
|
elif opt_lower == 'rmsprop':
|
|
optimizer = optim.RMSprop(parameters, alpha=0.9, momentum=args.momentum, **opt_args)
|
|
else:
|
|
assert False and "Invalid optimizer"
|
|
raise ValueError
|
|
|
|
return optimizer
|
|
|
|
|
|
class SmoothedValue(object):
|
|
"""Track a series of values and provide access to smoothed values over a
|
|
window or the global series average.
|
|
"""
|
|
|
|
def __init__(self, window_size=20, fmt=None):
|
|
if fmt is None:
|
|
fmt = "{median:.4f} ({global_avg:.4f})"
|
|
self.deque = deque(maxlen=window_size)
|
|
self.total = 0.0
|
|
self.count = 0
|
|
self.fmt = fmt
|
|
|
|
def update(self, value, n=1):
|
|
self.deque.append(value)
|
|
self.count += n
|
|
self.total += value * n
|
|
|
|
def synchronize_between_processes(self):
|
|
"""
|
|
Warning: does not synchronize the deque!
|
|
"""
|
|
if not is_dist_avail_and_initialized():
|
|
return
|
|
t = torch.tensor([self.count, self.total], dtype=torch.float64, device='cuda')
|
|
dist.barrier()
|
|
dist.all_reduce(t)
|
|
t = t.tolist()
|
|
self.count = int(t[0])
|
|
self.total = t[1]
|
|
|
|
@property
|
|
def median(self):
|
|
d = torch.tensor(list(self.deque))
|
|
return d.median().item()
|
|
|
|
@property
|
|
def avg(self):
|
|
d = torch.tensor(list(self.deque), dtype=torch.float32)
|
|
return d.mean().item()
|
|
|
|
@property
|
|
def global_avg(self):
|
|
return self.total / self.count
|
|
|
|
@property
|
|
def max(self):
|
|
return max(self.deque)
|
|
|
|
@property
|
|
def value(self):
|
|
return self.deque[-1]
|
|
|
|
def __str__(self):
|
|
return self.fmt.format(
|
|
median=self.median,
|
|
avg=self.avg,
|
|
global_avg=self.global_avg,
|
|
max=self.max,
|
|
value=self.value)
|
|
|
|
|
|
class MetricLogger(object):
|
|
def __init__(self, delimiter="\t"):
|
|
self.meters = defaultdict(SmoothedValue)
|
|
self.delimiter = delimiter
|
|
|
|
def update(self, **kwargs):
|
|
for k, v in kwargs.items():
|
|
if v is None:
|
|
continue
|
|
if isinstance(v, torch.Tensor):
|
|
v = v.item()
|
|
assert isinstance(v, (float, int))
|
|
self.meters[k].update(v)
|
|
|
|
def __getattr__(self, attr):
|
|
if attr in self.meters:
|
|
return self.meters[attr]
|
|
if attr in self.__dict__:
|
|
return self.__dict__[attr]
|
|
raise AttributeError("'{}' object has no attribute '{}'".format(
|
|
type(self).__name__, attr))
|
|
|
|
def __str__(self):
|
|
loss_str = []
|
|
for name, meter in self.meters.items():
|
|
loss_str.append(
|
|
"{}: {}".format(name, str(meter))
|
|
)
|
|
return self.delimiter.join(loss_str)
|
|
|
|
def synchronize_between_processes(self):
|
|
for meter in self.meters.values():
|
|
meter.synchronize_between_processes()
|
|
|
|
def add_meter(self, name, meter):
|
|
self.meters[name] = meter
|
|
|
|
def log_every(self, iterable, print_freq, header=None):
|
|
i = 0
|
|
if not header:
|
|
header = ''
|
|
start_time = time.time()
|
|
end = time.time()
|
|
iter_time = SmoothedValue(fmt='{avg:.4f}')
|
|
data_time = SmoothedValue(fmt='{avg:.4f}')
|
|
space_fmt = ':' + str(len(str(len(iterable)))) + 'd'
|
|
log_msg = [
|
|
header,
|
|
'[{0' + space_fmt + '}/{1}]',
|
|
'eta: {eta}',
|
|
'{meters}',
|
|
'time: {time}',
|
|
'data: {data}'
|
|
]
|
|
if torch.cuda.is_available():
|
|
log_msg.append('max mem: {memory:.0f}')
|
|
log_msg = self.delimiter.join(log_msg)
|
|
MB = 1024.0 * 1024.0
|
|
for obj in iterable:
|
|
data_time.update(time.time() - end)
|
|
yield obj
|
|
iter_time.update(time.time() - end)
|
|
if i % print_freq == 0 or i == len(iterable) - 1:
|
|
eta_seconds = iter_time.global_avg * (len(iterable) - i)
|
|
eta_string = str(datetime.timedelta(seconds=int(eta_seconds)))
|
|
if torch.cuda.is_available():
|
|
print(log_msg.format(
|
|
i, len(iterable), eta=eta_string,
|
|
meters=str(self),
|
|
time=str(iter_time), data=str(data_time),
|
|
memory=torch.cuda.max_memory_allocated() / MB))
|
|
else:
|
|
print(log_msg.format(
|
|
i, len(iterable), eta=eta_string,
|
|
meters=str(self),
|
|
time=str(iter_time), data=str(data_time)))
|
|
i += 1
|
|
end = time.time()
|
|
total_time = time.time() - start_time
|
|
total_time_str = str(datetime.timedelta(seconds=int(total_time)))
|
|
print('{} Total time: {} ({:.4f} s / it)'.format(
|
|
header, total_time_str, total_time / len(iterable))) |