utils.py

# --------------------------------------------------------
# Based on BEiT v2, SLIP code bases
# https://github.com/microsoft/unilm/tree/master/beit2
# https://github.com/facebookresearch/SLIP
# --------------------------------------------------------'
import numpy as np
import os
import random
import shutil
import torch
import torch.distributed as dist
import torch.autograd as autograd

from PIL import ImageFilter, ImageOps
from datetime import datetime

def unwrap_model(model):
    if hasattr(model, 'module'):
        return model.module
    else:
        return model

def get_model(model):
    if isinstance(model, torch.nn.DataParallel) \
      or isinstance(model, torch.nn.parallel.DistributedDataParallel):
        return model.module
    else:
        return model
    
class AverageMeter(object):
    """Computes and stores the average and current value"""
    def __init__(self, name, fmt=':f'):
        self.name = name
        self.fmt = fmt
        self.reset()

    def reset(self):
        self.val = 0
        self.avg = 0
        self.sum = 0
        self.count = 0

    def update(self, val, n=1):
        self.val = val
        self.sum += val * n
        self.count += n
        self.avg = self.sum / self.count

    def synchronize(self):
        if not is_dist_avail_and_initialized():
            return
        t = torch.tensor([self.sum, self.count], dtype=torch.float64, device='cuda')
        dist.barrier()
        dist.all_reduce(t)
        t = t.tolist()
        self.sum = int(t[0])
        self.count = t[1]
        self.avg = self.sum / self.count

    def __str__(self):
        fmtstr = '{name} {val' + self.fmt + '} ({avg' + self.fmt + '})'
        return fmtstr.format(**self.__dict__)


class ProgressMeter(object):
    def __init__(self, num_batches, meters, prefix=""):
        self.batch_fmtstr = self._get_batch_fmtstr(num_batches)
        self.meters = meters
        self.prefix = prefix

    def display(self, batch):
        entries = [self.prefix + self.batch_fmtstr.format(batch)]
        entries += [str(meter) for meter in self.meters]
        print('\t'.join(entries))

    def synchronize(self):
        for meter in self.meters:
            meter.synchronize()

    def _get_batch_fmtstr(self, num_batches):
        num_digits = len(str(num_batches // 1))
        fmt = '{:' + str(num_digits) + 'd}'
        return '[' + fmt + '/' + fmt.format(num_batches) + ']'


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.reshape(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 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(state, is_best, output_dir,epoch=None,toal_epoch=25):
    if is_main_process():
        ckpt_path = f'{output_dir}/checkpoint.pt'
        best_path = f'{output_dir}/checkpoint_best.pt'
        torch.save(state, ckpt_path)
        if epoch is not None:
            save_frep = toal_epoch*0.2
            if epoch % save_frep == 0:
                each_ckpt_path = f'{output_dir}/checkpoint_{epoch}.pt'
                shutil.copyfile(ckpt_path, each_ckpt_path)
        if is_best:
            shutil.copyfile(ckpt_path, best_path)


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'])
    elif 'SLURM_PROCID' in os.environ:
        args.rank = int(os.environ['SLURM_PROCID'])
        args.gpu = args.rank % torch.cuda.device_count()
    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)


def scaled_all_reduce(tensors, is_scale=True):
    """Performs the scaled all_reduce operation on the provided tensors.
    The input tensors are modified in-place. Currently supports only the sum
    reduction operator. The reduced values are scaled by the inverse size of the
    world size.
    """
    world_size = get_world_size()
    # There is no need for reduction in the single-proc case
    if world_size == 1:
        return tensors
    # Queue the reductions
    reductions = []
    for tensor in tensors:
        reduction = dist.all_reduce(tensor, async_op=True)
        reductions.append(reduction)
    # Wait for reductions to finish
    for reduction in reductions:
        reduction.wait()
    # Scale the results
    if is_scale:
        for tensor in tensors:
            tensor.mul_(1.0 / world_size)
    return tensors


class GatherLayer(autograd.Function):
    """
    Gather tensors from all workers with support for backward propagation:
    This implementation does not cut the gradients as torch.distributed.all_gather does.
    """

    @staticmethod
    def forward(ctx, x):
        output = [torch.zeros_like(x) for _ in range(dist.get_world_size())]
        dist.all_gather(output, x)
        return tuple(output)

    @staticmethod
    def backward(ctx, *grads):
        all_gradients = torch.stack(grads)
        dist.all_reduce(all_gradients)
        return all_gradients[dist.get_rank()]


def all_gather_batch_with_grad(tensors):
    """
    Performs all_gather operation on the provided tensors.
    Graph remains connected for backward grad computation.
    """
    # Queue the gathered tensors
    world_size = get_world_size()
    # There is no need for reduction in the single-proc case
    if world_size == 1:
        return tensors
    tensor_list = []
    output_tensor = []

    for tensor in tensors:
        tensor_all = GatherLayer.apply(tensor)
        tensor_list.append(tensor_all)

    for tensor_all in tensor_list:
        output_tensor.append(torch.cat(tensor_all, dim=0))
    return output_tensor


def cosine_scheduler(base_value, final_value, epochs, niter_per_ep, warmup_epochs=0, start_warmup_value=0):
    warmup_schedule = np.array([])
    warmup_iters = warmup_epochs * niter_per_ep
    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 = final_value + 0.5 * (base_value - final_value) * (1 + np.cos(np.pi * iters / len(iters)))

    schedule = np.concatenate((warmup_schedule, schedule))
    assert len(schedule) == epochs * niter_per_ep
    return schedule


class GaussianBlur(object):
    """Gaussian blur augmentation in SimCLR https://arxiv.org/abs/2002.05709"""

    def __init__(self, sigma=[.1, 2.]):
        self.sigma = sigma

    def __call__(self, x):
        sigma = random.uniform(self.sigma[0], self.sigma[1])
        x = x.filter(ImageFilter.GaussianBlur(radius=sigma))
        return x
class Solarize(object):
    """Solarize augmentation from BYOL: https://arxiv.org/abs/2006.07733"""

    def __call__(self, x):
        return ImageOps.solarize(x)

# from https://github.com/facebookresearch/mae/blob/main/util/pos_embed.py
def get_2d_sincos_pos_embed(embed_dim, grid_size, cls_token=False):
    """
    grid_size: int of the grid height and width
    return:
    pos_embed: [grid_size*grid_size, embed_dim] or [1+grid_size*grid_size, embed_dim] (w/ or w/o cls_token)
    """
    grid_h = np.arange(grid_size, dtype=np.float32)
    grid_w = np.arange(grid_size, dtype=np.float32)
    grid = np.meshgrid(grid_w, grid_h)  # here w goes first
    grid = np.stack(grid, axis=0)

    grid = grid.reshape([2, 1, grid_size, grid_size])
    pos_embed = get_2d_sincos_pos_embed_from_grid(embed_dim, grid)
    if cls_token:
        pos_embed = np.concatenate([np.zeros([1, embed_dim]), pos_embed], axis=0)
    return pos_embed


def get_2d_sincos_pos_embed_from_grid(embed_dim, grid):
    assert embed_dim % 2 == 0

    # use half of dimensions to encode grid_h
    emb_h = get_1d_sincos_pos_embed_from_grid(embed_dim // 2, grid[0])  # (H*W, D/2)
    emb_w = get_1d_sincos_pos_embed_from_grid(embed_dim // 2, grid[1])  # (H*W, D/2)

    emb = np.concatenate([emb_h, emb_w], axis=1) # (H*W, D)
    return emb


def get_1d_sincos_pos_embed_from_grid(embed_dim, pos):
    """
    embed_dim: output dimension for each position
    pos: a list of positions to be encoded: size (M,)
    out: (M, D)
    """
    assert embed_dim % 2 == 0
    omega = np.arange(embed_dim // 2, dtype=np.float)
    omega /= embed_dim / 2.
    omega = 1. / 10000**omega  # (D/2,)

    pos = pos.reshape(-1)  # (M,)
    out = np.einsum('m,d->md', pos, omega)  # (M, D/2), outer product

    emb_sin = np.sin(out) # (M, D/2)
    emb_cos = np.cos(out) # (M, D/2)

    emb = np.concatenate([emb_sin, emb_cos], axis=1)  # (M, D)
    return emb


# --------------------------------------------------------
# Interpolate position embeddings for high-resolution
# References:
# DeiT: https://github.com/facebookresearch/deit
# --------------------------------------------------------
def interpolate_pos_embed(model, checkpoint_model):
    if 'pos_embed' in checkpoint_model:
        pos_embed_checkpoint = checkpoint_model['pos_embed']
        embedding_size = pos_embed_checkpoint.shape[-1]
        num_patches = model.patch_embed.num_patches
        num_extra_tokens = model.pos_embed.shape[-2] - num_patches
        # height (== width) for the checkpoint position embedding
        orig_size = int((pos_embed_checkpoint.shape[-2] - num_extra_tokens) ** 0.5)
        # height (== width) for the new position embedding
        new_size = int(num_patches ** 0.5)
        # class_token and dist_token are kept unchanged
        if orig_size != new_size:
            print("Position interpolate from %dx%d to %dx%d" % (orig_size, orig_size, new_size, new_size))
            extra_tokens = pos_embed_checkpoint[:, :num_extra_tokens]
            # only the position tokens are interpolated
            pos_tokens = pos_embed_checkpoint[:, num_extra_tokens:]
            pos_tokens = pos_tokens.reshape(-1, orig_size, orig_size, embedding_size).permute(0, 3, 1, 2)
            pos_tokens = torch.nn.functional.interpolate(
                pos_tokens, size=(new_size, new_size), mode='bicubic', align_corners=False)
            pos_tokens = pos_tokens.permute(0, 2, 3, 1).flatten(1, 2)
            new_pos_embed = torch.cat((extra_tokens, pos_tokens), dim=1)
            checkpoint_model['pos_embed'] = new_pos_embed