util/misc.py

"""
Misc functions, including distributed helpers.
Mostly copy-paste from torchvision references.
"""
import os
import io
import subprocess
import time
from collections import defaultdict, deque
import datetime
from typing import Optional, List, Dict, Any

import torch
import torch.distributed as dist
from torch import Tensor
import functools


_LOCAL_PROCESS_GROUP = None

@functools.lru_cache()
def _get_global_gloo_group():
    """
    Return a process group based on gloo backend, containing all the ranks
    The result is cached.
    """

    if dist.get_backend() == "nccl":
        return dist.new_group(backend="gloo")

    return dist.group.WORLD


# needed due to empty tensor bug in pytorch and torchvision 0.5
import torchvision
# if float(torchvision.__version__[:3]) < 0.7:
#     from torchvision.ops import _new_empty_tensor
#     from torchvision.ops.misc import _output_size


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)

# copy-paste from mdetr: https://github.com/ashkamath/mdetr/blob/main/util/dist.py
def all_gather(data):
    """
    Run all_gather on arbitrary picklable data (not necessarily tensors)
    Args:
        data: any picklable object
    Returns:
        list[data]: list of data gathered from each rank
    """

    world_size = get_world_size()
    if world_size == 1:
        return [data]

    cpu_group = None
    if os.getenv("MDETR_CPU_REDUCE") == "1":
        cpu_group = _get_global_gloo_group()

    buffer = io.BytesIO()
    torch.save(data, buffer)
    data_view = buffer.getbuffer()
    device = "cuda" if cpu_group is None else "cpu"
    tensor = torch.ByteTensor(data_view).to(device)

    # obtain Tensor size of each rank
    local_size = torch.tensor([tensor.numel()], device=device, dtype=torch.long)
    size_list = [torch.tensor([0], device=device, dtype=torch.long) for _ in range(world_size)]
    if cpu_group is None:
        dist.all_gather(size_list, local_size)
    else:
        print("gathering on cpu")
        dist.all_gather(size_list, local_size, group=cpu_group)
    size_list = [int(size.item()) for size in size_list]
    max_size = max(size_list)
    assert isinstance(local_size.item(), int)
    local_size = int(local_size.item())

    # receiving Tensor from all ranks
    # we pad the tensor because torch all_gather does not support
    # gathering tensors of different shapes
    tensor_list = []
    for _ in size_list:
        tensor_list.append(torch.empty((max_size,), dtype=torch.uint8, device=device))
    if local_size != max_size:
        padding = torch.empty(size=(max_size - local_size,), dtype=torch.uint8, device=device)
        tensor = torch.cat((tensor, padding), dim=0)
    if cpu_group is None:
        dist.all_gather(tensor_list, tensor)
    else:
        dist.all_gather(tensor_list, tensor, group=cpu_group)

    data_list = []
    for size, tensor in zip(size_list, tensor_list):
        tensor = torch.split(tensor, [size, max_size - size], dim=0)[0]
        buffer = io.BytesIO(tensor.cpu().numpy())
        obj = torch.load(buffer)
        data_list.append(obj)

    return data_list

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 = 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)
        dist.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:.4f}')
        data_time = SmoothedValue(fmt='{avg:.4f}')
        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: {} ({:.4f} s / it)'.format(
            header, total_time_str, total_time / len(iterable)))


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
    message = f"sha: {sha}, status: {diff}, branch: {branch}"
    return message


def collate_fn(batch):
    # batch: imgs, targets
    batch = list(zip(*batch)) 
    batch[0] = nested_tensor_from_videos_list(batch[0], size_divisibility=32) 
    # batch[0]: samples: NestedTensor(tensor, mask)
    #           tensor: [B, T, C, H, W], mask: [B, T, H, W]
    # batch[1]: targets: list[dict]
    return tuple(batch) 


def _max_by_axis(the_list):
    # type: (List[List[int]]) -> List[int]
    maxes = the_list[0]
    for sublist in the_list[1:]: # (C, H, W)
        for index, item in enumerate(sublist):
            maxes[index] = max(maxes[index], item)
    return maxes


def nested_tensor_from_tensor_list(tensor_list: List[Tensor], size_divisibility=1, split=True):
    """
    This function receives a list of image tensors and returns a NestedTensor of the padded images, along with their
    padding masks (true for padding areas, false otherwise).
    """
    # TODO make this more general
    # if image tensor is stacked as [T*3, H, W], then use split
    if split:
        tensor_list = [tensor.split(3,dim=0) for tensor in tensor_list] 
        tensor_list = [item for sublist in tensor_list for item in sublist] 
        # list[tensor], length = batch_size x time

    if tensor_list[0].ndim == 3: 
        # TODO make it support different-sized images
        max_size = _max_by_axis([list(img.shape) for img in tensor_list]) 

        if size_divisibility > 1: # so that the mask dowmsample can be matched
            stride = size_divisibility
            # the last two dims are [H, W], both subject to divisibility requirement
            max_size[-2] = (max_size[-2] + (stride - 1)) // stride * stride
            max_size[-1] = (max_size[-1] + (stride - 1)) // stride * stride

        # min_size = tuple(min(s) for s in zip(*[img.shape for img in tensor_list]))
        batch_shape = [len(tensor_list)] + max_size 
        b, c, h, w = batch_shape 
        dtype = tensor_list[0].dtype
        device = tensor_list[0].device
        tensor = torch.zeros(batch_shape, dtype=dtype, device=device) 
        mask = torch.ones((b, h, w), dtype=torch.bool, device=device)
        for img, pad_img, m in zip(tensor_list, tensor, mask):
            pad_img[: img.shape[0], : img.shape[1], : img.shape[2]].copy_(img)
            m[: img.shape[1], :img.shape[2]] = False # valid locations
    else:
        raise ValueError('not supported')
    return NestedTensor(tensor, mask)

def nested_tensor_from_videos_list(videos_list: List[Tensor], size_divisibility=1):
    """
    This function receives a list of videos (each of shape [T, C, H, W]) and returns a NestedTensor of the padded
    videos (shape [B, T, C, PH, PW], along with their padding masks (true for padding areas, false otherwise, of shape
    [B, T, PH, PW].
    """
    max_size = _max_by_axis([list(img.shape) for img in videos_list])

    if size_divisibility > 1: # so that the mask dowmsample can be matched
        stride = size_divisibility
        # the last two dims are [H, W], both subject to divisibility requirement
        max_size[-2] = (max_size[-2] + (stride - 1)) // stride * stride
        max_size[-1] = (max_size[-1] + (stride - 1)) // stride * stride

    padded_batch_shape = [len(videos_list)] + max_size
    b, t, c, h, w = padded_batch_shape
    dtype = videos_list[0].dtype
    device = videos_list[0].device
    padded_videos = torch.zeros(padded_batch_shape, dtype=dtype, device=device)
    videos_pad_masks = torch.ones((b, t, h, w), dtype=torch.bool, device=device)
    for vid_frames, pad_vid_frames, vid_pad_m in zip(videos_list, padded_videos, videos_pad_masks):
        pad_vid_frames[:vid_frames.shape[0], :, :vid_frames.shape[2], :vid_frames.shape[3]].copy_(vid_frames)
        vid_pad_m[:vid_frames.shape[0], :vid_frames.shape[2], :vid_frames.shape[3]] = False
    return NestedTensor(padded_videos, videos_pad_masks)


class NestedTensor(object):
    def __init__(self, tensors, mask: Optional[Tensor]):
        self.tensors = tensors
        self.mask = mask

    def to(self, device):
        # type: (Device) -> NestedTensor # noqa
        cast_tensor = self.tensors.to(device)
        mask = self.mask
        if mask is not None:
            assert mask is not None
            cast_mask = mask.to(device)
        else:
            cast_mask = None
        return NestedTensor(cast_tensor, cast_mask)

    def decompose(self):
        return self.tensors, self.mask

    def __repr__(self):
        return str(self.tensors)


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 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, _use_new_zipfile_serialization=False)


def init_distributed_mode(args):
    if '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'])
        args.dist_url = 'env://'
        os.environ['LOCAL_SIZE'] = str(torch.cuda.device_count())
    elif 'SLURM_PROCID' in os.environ:
        proc_id = int(os.environ['SLURM_PROCID'])
        ntasks = int(os.environ['SLURM_NTASKS'])
        node_list = os.environ['SLURM_NODELIST']
        num_gpus = torch.cuda.device_count()
        addr = subprocess.getoutput(
            'scontrol show hostname {} | head -n1'.format(node_list))
        os.environ['MASTER_PORT'] = os.environ.get('MASTER_PORT', '29500')
        os.environ['MASTER_ADDR'] = addr
        os.environ['WORLD_SIZE'] = str(ntasks)
        os.environ['RANK'] = str(proc_id)
        os.environ['LOCAL_RANK'] = str(proc_id % num_gpus)
        os.environ['LOCAL_SIZE'] = str(num_gpus)
        args.dist_url = 'env://'
        args.world_size = ntasks
        args.rank = proc_id
        args.gpu = proc_id % num_gpus
    else:
        print('Not using distributed mode')
        args.distributed = False
        return

    args.distributed = True

    torch.cuda.set_device(args.gpu)
    args.dist_backend = 'nccl'
    print('| distributed init (rank {}): {}'.format(
        args.rank, args.dist_url), flush=True)
    torch.distributed.init_process_group(backend=args.dist_backend, init_method=args.dist_url,
                                         world_size=args.world_size, rank=args.rank)
    torch.distributed.barrier()
    setup_for_distributed(args.rank == 0)


@torch.no_grad()
def accuracy(output, target, topk=(1,)):
    """Computes the precision@k for the specified values of k"""
    if target.numel() == 0:
        return [torch.zeros([], device=output.device)]
    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].view(-1).float().sum(0)
        res.append(correct_k.mul_(100.0 / batch_size))
    return res


def interpolate(input, size=None, scale_factor=None, mode="nearest", align_corners=None):
    # type: (Tensor, Optional[List[int]], Optional[float], str, Optional[bool]) -> Tensor
    """
    Equivalent to nn.functional.interpolate, but with support for empty batch sizes.
    This will eventually be supported natively by PyTorch, and this
    class can go away.
    """
    # if float(torchvision.__version__[:3]) < 0.7:
    #     if input.numel() > 0:
    #         return torch.nn.functional.interpolate(
    #             input, size, scale_factor, mode, align_corners
    #         )

    #     output_shape = _output_size(2, input, size, scale_factor)
    #     output_shape = list(input.shape[:-2]) + list(output_shape)
    #     return _new_empty_tensor(input, output_shape)
    # else:
    #     return torchvision.ops.misc.interpolate(input, size, scale_factor, mode, align_corners)
    return torchvision.ops.misc.interpolate(input, size, scale_factor, mode, align_corners)


def targets_to(targets: List[Dict[str, Any]], device):
    """Moves the target dicts to the given device.
    excluded_keys = [
        "questionId",
        "tokens_positive",
        "tokens",
        "dataset_name",
        "sentence_id",
        "original_img_id",
        "nb_eval",
        "task_id",
        "original_id",
    ]
    """
    if "dataset_name" in targets[0]:
        # for ["refcoco", "refcoco+", "refcocog"] evaluation
        return [{k: v.to(device)  for k, v in t.items() if k not in ["caption", "dataset_name", "original_id"]} for t in targets]
    else:
        return [{k: v.to(device)  for k, v in t.items() if k not in ["caption", "dataset_name", "original_id", "image_id"]} for t in targets]

def get_total_grad_norm(parameters, norm_type=2):
    parameters = list(filter(lambda p: p.grad is not None, parameters))
    norm_type = float(norm_type)
    device = parameters[0].grad.device
    total_norm = torch.norm(torch.stack([torch.norm(p.grad.detach(), norm_type).to(device) for p in parameters]),
                            norm_type)



def inverse_sigmoid(x, eps=1e-5):
    x = x.clamp(min=0, max=1)
    x1 = x.clamp(min=eps)
    x2 = (1 - x).clamp(min=eps)
    return torch.log(x1/x2)