utils.py

import argparse
import datetime
import os
import subprocess
import time
from collections import defaultdict, deque

import numpy as np
import torch
import torch.nn as nn
import torch.distributed

from utils import dist as dist


def fix_random_seeds(seed=0):
    torch.manual_seed(seed)
    torch.cuda.manual_seed_all(seed)
    np.random.seed(seed)


def get_sha():
    cwd = os.path.dirname(os.path.abspath(__file__))

    def _run(command):
        return subprocess.check_output(command, cwd=cwd).decode('ascii').strip()

    sha = 'N/A'
    diff = "clean"
    branch = 'N/A'
    try:
        sha = _run(['git', 'rev-parse', 'HEAD'])
        subprocess.check_output(['git', 'diff'], cwd=cwd)
        diff = _run(['git', 'diff-index', 'HEAD'])
        diff = "has uncommited changes" if diff else "clean"
        branch = _run(['git', 'rev-parse', '--abbrev-ref', 'HEAD'])
    except Exception:
        pass
    return f"sha: {sha}, status: {diff}, branch: {branch}"


def has_batchnorms(model):
    bn_types = (nn.BatchNorm1d, nn.BatchNorm2d, nn.BatchNorm3d, nn.SyncBatchNorm)
    for name, module in model.named_modules():
        if isinstance(module, bn_types):
            return True
    return False


def restart_from_checkpoint(ckp_path, run_variables=None, **kwargs):
    """
    Re-start from checkpoint
    """
    if not os.path.isfile(ckp_path):
        return
    print("Found checkpoint at {}".format(ckp_path))

    # open checkpoint file
    checkpoint = torch.load(ckp_path, map_location="cpu")

    # key is what to look for in the checkpoint file
    # value is the object to load
    # example: {'state_dict': model}
    for key, value in kwargs.items():
        if key in checkpoint and value is not None:
            try:
                msg = value.load_state_dict(checkpoint[key], strict=False)
                print("=> loaded '{}' from checkpoint '{}' with msg {}".format(key, ckp_path, msg))
            except TypeError:
                try:
                    msg = value.load_state_dict(checkpoint[key])
                    print("=> loaded '{}' from checkpoint: '{}'".format(key, ckp_path))
                except ValueError:
                    print("=> failed to load '{}' from checkpoint: '{}'".format(key, ckp_path))
        else:
            print("=> key '{}' not found in checkpoint: '{}'".format(key, ckp_path))

    # re load variable important for the run
    if run_variables is not None:
        for var_name in run_variables:
            if var_name in checkpoint:
                run_variables[var_name] = checkpoint[var_name]


def save_on_master(*args, **kwargs):
    if dist.is_main_process():
        torch.save(*args, **kwargs)


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:.6f} ({global_avg:.6f})"
        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 dist.is_enabled():
            return
        t = torch.tensor([self.count, self.total], dtype=torch.float64, device='cuda')
        torch.distributed.barrier()
        torch.distributed.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)


def reduce_dict(input_dict, average=True):
    """
    Args:
        input_dict (dict): all the values will be reduced
        average (bool): whether to do average or sum
    Reduce the values in the dictionary from all processes so that all processes
    have the averaged results. Returns a dict with the same fields as
    input_dict, after reduction.
    """
    world_size = dist.get_world_size()
    if world_size < 2:
        return input_dict
    with torch.no_grad():
        names = []
        values = []
        # sort the keys so that they are consistent across processes
        for k in sorted(input_dict.keys()):
            names.append(k)
            values.append(input_dict[k])
        values = torch.stack(values, dim=0)
        torch.distributed.all_reduce(values)
        if average:
            values /= world_size
        reduced_dict = {k: v for k, v in zip(names, values)}
    return reduced_dict


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 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:.6f}')
        data_time = SmoothedValue(fmt='{avg:.6f}')
        space_fmt = ':' + str(len(str(len(iterable)))) + 'd'
        if torch.cuda.is_available():
            log_msg = self.delimiter.join([
                header,
                '[{0' + space_fmt + '}/{1}]',
                'eta: {eta}',
                '{meters}',
                'time: {time}',
                'data: {data}',
                'max mem: {memory:.0f}'
            ])
        else:
            log_msg = self.delimiter.join([
                header,
                '[{0' + space_fmt + '}/{1}]',
                'eta: {eta}',
                '{meters}',
                'time: {time}',
                'data: {data}'
            ])
        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: {} ({:.6f} s / it)'.format(
            header, total_time_str, total_time / len(iterable)))


def load_pretrained_weights(model, pretrained_weights, checkpoint_key):
    if os.path.isfile(pretrained_weights):
        state_dict = torch.load(pretrained_weights, map_location="cpu")
        if checkpoint_key is not None and checkpoint_key in state_dict:
            print(f"Take key {checkpoint_key} in provided checkpoint dict")
            state_dict = state_dict[checkpoint_key]
        for k in list(state_dict.keys()):
            # retain only encoder up to before the embedding layer
            if k.startswith('module.encoder') and not k.startswith('module.encoder.fc'):
                # remove prefix
                state_dict[k[len("module.encoder."):]] = state_dict[k]
            # delete renamed or unused k
            del state_dict[k]

        msg = model.load_state_dict(state_dict, strict=False)
        assert set(msg.missing_keys) == {"fc.weight", "fc.bias"}

        print('Pretrained weights found at {} and loaded with msg: {}'.format(pretrained_weights, msg))
    else:
        print("=> no checkpoint found at '{}'".format(pretrained_weights))


def accuracy(output, target, topk=(1,)):
    """Computes the accuracy over the k top predictions for the specified values of k"""
    with torch.no_grad():
        maxk = max(topk)
        batch_size = target.size(0)

        _, pred = output.topk(maxk, 1, True, True)
        pred = pred.t()
        correct = pred.eq(target.view(1, -1).expand_as(pred))

        res = []
        for k in topk:
            correct_k = correct[:k].reshape(-1).float().sum(0, keepdim=True)
            res.append(correct_k.mul_(100.0 / batch_size))
        return res


def bool_flag(s):
    """
    Parse boolean arguments from the command line.
    """
    FALSY_STRINGS = {"off", "false", "0"}
    TRUTHY_STRINGS = {"on", "true", "1"}
    if s.lower() in FALSY_STRINGS:
        return False
    elif s.lower() in TRUTHY_STRINGS:
        return True
    else:
        raise argparse.ArgumentTypeError("invalid value for a boolean flag")


def find_free_port():
    import socket
    from contextlib import closing

    with closing(socket.socket(socket.AF_INET, socket.SOCK_STREAM)) as s:
        s.bind(("", 0))
        s.setsockopt(socket.SOL_SOCKET, socket.SO_REUSEADDR, 1)
        return s.getsockname()[1]