增加VIT模型代码

PyTorch/NLP/Vision_Transformer/submitit_finetune.py

+# Copyright (c) Meta Platforms, Inc. and affiliates.
+# All rights reserved.
+
+# This source code is licensed under the license found in the
+# LICENSE file in the root directory of this source tree.
+# --------------------------------------------------------
+# A script to run multinode training with submitit.
+# --------------------------------------------------------
+
+import argparse
+import os
+import uuid
+from pathlib import Path
+
+import main_finetune as classification
+import submitit
+
+
+def parse_args():
+    classification_parser = classification.get_args_parser()
+    parser = argparse.ArgumentParser("Submitit for MAE finetune", parents=[classification_parser])
+    parser.add_argument("--ngpus", default=8, type=int, help="Number of gpus to request on each node")
+    parser.add_argument("--nodes", default=2, type=int, help="Number of nodes to request")
+    parser.add_argument("--timeout", default=4320, type=int, help="Duration of the job")
+    parser.add_argument("--job_dir", default="", type=str, help="Job dir. Leave empty for automatic.")
+
+    parser.add_argument("--partition", default="learnfair", type=str, help="Partition where to submit")
+    parser.add_argument("--use_volta32", action='store_true', help="Request 32G V100 GPUs")
+    parser.add_argument('--comment', default="", type=str, help="Comment to pass to scheduler")
+    return parser.parse_args()
+
+
+def get_shared_folder() -> Path:
+    user = os.getenv("USER")
+    if Path("/checkpoint/").is_dir():
+        p = Path(f"/checkpoint/{user}/experiments")
+        p.mkdir(exist_ok=True)
+        return p
+    raise RuntimeError("No shared folder available")
+
+
+def get_init_file():
+    # Init file must not exist, but it's parent dir must exist.
+    os.makedirs(str(get_shared_folder()), exist_ok=True)
+    init_file = get_shared_folder() / f"{uuid.uuid4().hex}_init"
+    if init_file.exists():
+        os.remove(str(init_file))
+    return init_file
+
+
+class Trainer(object):
+    def __init__(self, args):
+        self.args = args
+
+    def __call__(self):
+        import main_finetune as classification
+
+        self._setup_gpu_args()
+        classification.main(self.args)
+
+    def checkpoint(self):
+        import os
+        import submitit
+
+        self.args.dist_url = get_init_file().as_uri()
+        checkpoint_file = os.path.join(self.args.output_dir, "checkpoint.pth")
+        if os.path.exists(checkpoint_file):
+            self.args.resume = checkpoint_file
+        print("Requeuing ", self.args)
+        empty_trainer = type(self)(self.args)
+        return submitit.helpers.DelayedSubmission(empty_trainer)
+
+    def _setup_gpu_args(self):
+        import submitit
+        from pathlib import Path
+
+        job_env = submitit.JobEnvironment()
+        self.args.output_dir = Path(str(self.args.output_dir).replace("%j", str(job_env.job_id)))
+        self.args.log_dir = self.args.output_dir
+        self.args.gpu = job_env.local_rank
+        self.args.rank = job_env.global_rank
+        self.args.world_size = job_env.num_tasks
+        print(f"Process group: {job_env.num_tasks} tasks, rank: {job_env.global_rank}")
+
+
+def main():
+    args = parse_args()
+    if args.job_dir == "":
+        args.job_dir = get_shared_folder() / "%j"
+
+    # Note that the folder will depend on the job_id, to easily track experiments
+    executor = submitit.AutoExecutor(folder=args.job_dir, slurm_max_num_timeout=30)
+
+    num_gpus_per_node = args.ngpus
+    nodes = args.nodes
+    timeout_min = args.timeout
+
+    partition = args.partition
+    kwargs = {}
+    if args.use_volta32:
+        kwargs['slurm_constraint'] = 'volta32gb'
+    if args.comment:
+        kwargs['slurm_comment'] = args.comment
+
+    executor.update_parameters(
+        mem_gb=40 * num_gpus_per_node,
+        gpus_per_node=num_gpus_per_node,
+        tasks_per_node=num_gpus_per_node, # one task per GPU
+        cpus_per_task=10,
+        nodes=nodes,
+        timeout_min=timeout_min,
+        # Below are cluster dependent parameters
+        slurm_partition=partition,
+        slurm_signal_delay_s=120,
+        **kwargs
+    )
+
+    executor.update_parameters(name="mae")
+
+    args.dist_url = get_init_file().as_uri()
+    args.output_dir = args.job_dir
+
+    trainer = Trainer(args)
+    job = executor.submit(trainer)
+
+    # print("Submitted job_id:", job.job_id)
+    print(job.job_id)
+
+
+if __name__ == "__main__":
+    main()

PyTorch/NLP/Vision_Transformer/submitit_linprobe.py

+# Copyright (c) Meta Platforms, Inc. and affiliates.
+# All rights reserved.
+
+# This source code is licensed under the license found in the
+# LICENSE file in the root directory of this source tree.
+# --------------------------------------------------------
+# A script to run multinode training with submitit.
+# --------------------------------------------------------
+
+import argparse
+import os
+import uuid
+from pathlib import Path
+
+import main_linprobe as classification
+import submitit
+
+
+def parse_args():
+    classification_parser = classification.get_args_parser()
+    parser = argparse.ArgumentParser("Submitit for MAE linear probe", parents=[classification_parser])
+    parser.add_argument("--ngpus", default=8, type=int, help="Number of gpus to request on each node")
+    parser.add_argument("--nodes", default=2, type=int, help="Number of nodes to request")
+    parser.add_argument("--timeout", default=4320, type=int, help="Duration of the job")
+    parser.add_argument("--job_dir", default="", type=str, help="Job dir. Leave empty for automatic.")
+
+    parser.add_argument("--partition", default="learnfair", type=str, help="Partition where to submit")
+    parser.add_argument("--use_volta32", action='store_true', help="Request 32G V100 GPUs")
+    parser.add_argument('--comment', default="", type=str, help="Comment to pass to scheduler")
+    return parser.parse_args()
+
+
+def get_shared_folder() -> Path:
+    user = os.getenv("USER")
+    if Path("/checkpoint/").is_dir():
+        p = Path(f"/checkpoint/{user}/experiments")
+        p.mkdir(exist_ok=True)
+        return p
+    raise RuntimeError("No shared folder available")
+
+
+def get_init_file():
+    # Init file must not exist, but it's parent dir must exist.
+    os.makedirs(str(get_shared_folder()), exist_ok=True)
+    init_file = get_shared_folder() / f"{uuid.uuid4().hex}_init"
+    if init_file.exists():
+        os.remove(str(init_file))
+    return init_file
+
+
+class Trainer(object):
+    def __init__(self, args):
+        self.args = args
+
+    def __call__(self):
+        import main_linprobe as classification
+
+        self._setup_gpu_args()
+        classification.main(self.args)
+
+    def checkpoint(self):
+        import os
+        import submitit
+
+        self.args.dist_url = get_init_file().as_uri()
+        checkpoint_file = os.path.join(self.args.output_dir, "checkpoint.pth")
+        if os.path.exists(checkpoint_file):
+            self.args.resume = checkpoint_file
+        print("Requeuing ", self.args)
+        empty_trainer = type(self)(self.args)
+        return submitit.helpers.DelayedSubmission(empty_trainer)
+
+    def _setup_gpu_args(self):
+        import submitit
+        from pathlib import Path
+
+        job_env = submitit.JobEnvironment()
+        self.args.output_dir = Path(str(self.args.output_dir).replace("%j", str(job_env.job_id)))
+        self.args.log_dir = self.args.output_dir
+        self.args.gpu = job_env.local_rank
+        self.args.rank = job_env.global_rank
+        self.args.world_size = job_env.num_tasks
+        print(f"Process group: {job_env.num_tasks} tasks, rank: {job_env.global_rank}")
+
+
+def main():
+    args = parse_args()
+    if args.job_dir == "":
+        args.job_dir = get_shared_folder() / "%j"
+
+    # Note that the folder will depend on the job_id, to easily track experiments
+    executor = submitit.AutoExecutor(folder=args.job_dir, slurm_max_num_timeout=30)
+
+    num_gpus_per_node = args.ngpus
+    nodes = args.nodes
+    timeout_min = args.timeout
+
+    partition = args.partition
+    kwargs = {}
+    if args.use_volta32:
+        kwargs['slurm_constraint'] = 'volta32gb'
+    if args.comment:
+        kwargs['slurm_comment'] = args.comment
+
+    executor.update_parameters(
+        mem_gb=40 * num_gpus_per_node,
+        gpus_per_node=num_gpus_per_node,
+        tasks_per_node=num_gpus_per_node, # one task per GPU
+        cpus_per_task=10,
+        nodes=nodes,
+        timeout_min=timeout_min,
+        # Below are cluster dependent parameters
+        slurm_partition=partition,
+        slurm_signal_delay_s=120,
+        **kwargs
+    )
+
+    executor.update_parameters(name="mae")
+
+    args.dist_url = get_init_file().as_uri()
+    args.output_dir = args.job_dir
+
+    trainer = Trainer(args)
+    job = executor.submit(trainer)
+
+    # print("Submitted job_id:", job.job_id)
+    print(job.job_id)
+
+
+if __name__ == "__main__":
+    main()

PyTorch/NLP/Vision_Transformer/submitit_pretrain.py

+# Copyright (c) Meta Platforms, Inc. and affiliates.
+# All rights reserved.
+
+# This source code is licensed under the license found in the
+# LICENSE file in the root directory of this source tree.
+# --------------------------------------------------------
+# A script to run multinode training with submitit.
+# --------------------------------------------------------
+
+import argparse
+import os
+import uuid
+from pathlib import Path
+
+import main_pretrain as trainer
+import submitit
+
+
+def parse_args():
+    trainer_parser = trainer.get_args_parser()
+    parser = argparse.ArgumentParser("Submitit for MAE pretrain", parents=[trainer_parser])
+    parser.add_argument("--ngpus", default=8, type=int, help="Number of gpus to request on each node")
+    parser.add_argument("--nodes", default=2, type=int, help="Number of nodes to request")
+    parser.add_argument("--timeout", default=4320, type=int, help="Duration of the job")
+    parser.add_argument("--job_dir", default="", type=str, help="Job dir. Leave empty for automatic.")
+
+    parser.add_argument("--partition", default="learnfair", type=str, help="Partition where to submit")
+    parser.add_argument("--use_volta32", action='store_true', help="Request 32G V100 GPUs")
+    parser.add_argument('--comment', default="", type=str, help="Comment to pass to scheduler")
+    return parser.parse_args()
+
+
+def get_shared_folder() -> Path:
+    user = os.getenv("USER")
+    if Path("/checkpoint/").is_dir():
+        p = Path(f"/checkpoint/{user}/experiments")
+        p.mkdir(exist_ok=True)
+        return p
+    raise RuntimeError("No shared folder available")
+
+
+def get_init_file():
+    # Init file must not exist, but it's parent dir must exist.
+    os.makedirs(str(get_shared_folder()), exist_ok=True)
+    init_file = get_shared_folder() / f"{uuid.uuid4().hex}_init"
+    if init_file.exists():
+        os.remove(str(init_file))
+    return init_file
+
+
+class Trainer(object):
+    def __init__(self, args):
+        self.args = args
+
+    def __call__(self):
+        import main_pretrain as trainer
+
+        self._setup_gpu_args()
+        trainer.main(self.args)
+
+    def checkpoint(self):
+        import os
+        import submitit
+
+        self.args.dist_url = get_init_file().as_uri()
+        checkpoint_file = os.path.join(self.args.output_dir, "checkpoint.pth")
+        if os.path.exists(checkpoint_file):
+            self.args.resume = checkpoint_file
+        print("Requeuing ", self.args)
+        empty_trainer = type(self)(self.args)
+        return submitit.helpers.DelayedSubmission(empty_trainer)
+
+    def _setup_gpu_args(self):
+        import submitit
+        from pathlib import Path
+
+        job_env = submitit.JobEnvironment()
+        self.args.output_dir = Path(str(self.args.output_dir).replace("%j", str(job_env.job_id)))
+        self.args.log_dir = self.args.output_dir
+        self.args.gpu = job_env.local_rank
+        self.args.rank = job_env.global_rank
+        self.args.world_size = job_env.num_tasks
+        print(f"Process group: {job_env.num_tasks} tasks, rank: {job_env.global_rank}")
+
+
+def main():
+    args = parse_args()
+    if args.job_dir == "":
+        args.job_dir = get_shared_folder() / "%j"
+
+    # Note that the folder will depend on the job_id, to easily track experiments
+    executor = submitit.AutoExecutor(folder=args.job_dir, slurm_max_num_timeout=30)
+
+    num_gpus_per_node = args.ngpus
+    nodes = args.nodes
+    timeout_min = args.timeout
+
+    partition = args.partition
+    kwargs = {}
+    if args.use_volta32:
+        kwargs['slurm_constraint'] = 'volta32gb'
+    if args.comment:
+        kwargs['slurm_comment'] = args.comment
+
+    executor.update_parameters(
+        mem_gb=40 * num_gpus_per_node,
+        gpus_per_node=num_gpus_per_node,
+        tasks_per_node=num_gpus_per_node,  # one task per GPU
+        cpus_per_task=10,
+        nodes=nodes,
+        timeout_min=timeout_min,  # max is 60 * 72
+        # Below are cluster dependent parameters
+        slurm_partition=partition,
+        slurm_signal_delay_s=120,
+        **kwargs
+    )
+
+    executor.update_parameters(name="mae")
+
+    args.dist_url = get_init_file().as_uri()
+    args.output_dir = args.job_dir
+
+    trainer = Trainer(args)
+    job = executor.submit(trainer)
+
+    # print("Submitted job_id:", job.job_id)
+    print(job.job_id)
+
+
+if __name__ == "__main__":
+    main()

PyTorch/NLP/Vision_Transformer/util/crop.py

+# Copyright (c) Meta Platforms, Inc. and affiliates.
+# All rights reserved.
+
+# This source code is licensed under the license found in the
+# LICENSE file in the root directory of this source tree.
+
+import math
+
+import torch
+
+from torchvision import transforms
+from torchvision.transforms import functional as F
+
+
+class RandomResizedCrop(transforms.RandomResizedCrop):
+    """
+    RandomResizedCrop for matching TF/TPU implementation: no for-loop is used.
+    This may lead to results different with torchvision's version.
+    Following BYOL's TF code:
+    https://github.com/deepmind/deepmind-research/blob/master/byol/utils/dataset.py#L206
+    """
+    @staticmethod
+    def get_params(img, scale, ratio):
+        width, height = F._get_image_size(img)
+        area = height * width
+
+        target_area = area * torch.empty(1).uniform_(scale[0], scale[1]).item()
+        log_ratio = torch.log(torch.tensor(ratio))
+        aspect_ratio = torch.exp(
+            torch.empty(1).uniform_(log_ratio[0], log_ratio[1])
+        ).item()
+
+        w = int(round(math.sqrt(target_area * aspect_ratio)))
+        h = int(round(math.sqrt(target_area / aspect_ratio)))
+
+        w = min(w, width)
+        h = min(h, height)
+
+        i = torch.randint(0, height - h + 1, size=(1,)).item()
+        j = torch.randint(0, width - w + 1, size=(1,)).item()
+
+        return i, j, h, w
\ No newline at end of file

PyTorch/NLP/Vision_Transformer/util/datasets.py

+# Copyright (c) Meta Platforms, Inc. and affiliates.
+# All rights reserved.
+
+# This source code is licensed under the license found in the
+# LICENSE file in the root directory of this source tree.
+# --------------------------------------------------------
+# References:
+# DeiT: https://github.com/facebookresearch/deit
+# --------------------------------------------------------
+
+import os
+import PIL
+
+from torchvision import datasets, transforms
+
+from timm.data import create_transform
+from timm.data.constants import IMAGENET_DEFAULT_MEAN, IMAGENET_DEFAULT_STD
+
+
+def build_dataset(is_train, args):
+    transform = build_transform(is_train, args)
+
+    root = os.path.join(args.data_path, 'train' if is_train else 'val')
+    dataset = datasets.ImageFolder(root, transform=transform)
+
+    print(dataset)
+
+    return dataset
+
+
+def build_transform(is_train, args):
+    mean = IMAGENET_DEFAULT_MEAN
+    std = IMAGENET_DEFAULT_STD
+    # train transform
+    if is_train:
+        # this should always dispatch to transforms_imagenet_train
+        transform = create_transform(
+            input_size=args.input_size,
+            is_training=True,
+            color_jitter=args.color_jitter,
+            auto_augment=args.aa,
+            interpolation='bicubic',
+            re_prob=args.reprob,
+            re_mode=args.remode,
+            re_count=args.recount,
+            mean=mean,
+            std=std,
+        )
+        return transform
+
+    # eval transform
+    t = []
+    if args.input_size <= 224:
+        crop_pct = 224 / 256
+    else:
+        crop_pct = 1.0
+    size = int(args.input_size / crop_pct)
+    t.append(
+        transforms.Resize(size, interpolation=PIL.Image.BICUBIC),  # to maintain same ratio w.r.t. 224 images
+    )
+    t.append(transforms.CenterCrop(args.input_size))
+
+    t.append(transforms.ToTensor())
+    t.append(transforms.Normalize(mean, std))
+    return transforms.Compose(t)

PyTorch/NLP/Vision_Transformer/util/lars.py

+# Copyright (c) Meta Platforms, Inc. and affiliates.
+# All rights reserved.
+
+# This source code is licensed under the license found in the
+# LICENSE file in the root directory of this source tree.
+# --------------------------------------------------------
+# LARS optimizer, implementation from MoCo v3:
+# https://github.com/facebookresearch/moco-v3
+# --------------------------------------------------------
+
+import torch
+
+
+class LARS(torch.optim.Optimizer):
+    """
+    LARS optimizer, no rate scaling or weight decay for parameters <= 1D.
+    """
+    def __init__(self, params, lr=0, weight_decay=0, momentum=0.9, trust_coefficient=0.001):
+        defaults = dict(lr=lr, weight_decay=weight_decay, momentum=momentum, trust_coefficient=trust_coefficient)
+        super().__init__(params, defaults)
+
+    @torch.no_grad()
+    def step(self):
+        for g in self.param_groups:
+            for p in g['params']:
+                dp = p.grad
+
+                if dp is None:
+                    continue
+
+                if p.ndim > 1: # if not normalization gamma/beta or bias
+                    dp = dp.add(p, alpha=g['weight_decay'])
+                    param_norm = torch.norm(p)
+                    update_norm = torch.norm(dp)
+                    one = torch.ones_like(param_norm)
+                    q = torch.where(param_norm > 0.,
+                                    torch.where(update_norm > 0,
+                                    (g['trust_coefficient'] * param_norm / update_norm), one),
+                                    one)
+                    dp = dp.mul(q)
+
+                param_state = self.state[p]
+                if 'mu' not in param_state:
+                    param_state['mu'] = torch.zeros_like(p)
+                mu = param_state['mu']
+                mu.mul_(g['momentum']).add_(dp)
+                p.add_(mu, alpha=-g['lr'])
\ No newline at end of file

PyTorch/NLP/Vision_Transformer/util/lr_decay.py

+# Copyright (c) Meta Platforms, Inc. and affiliates.
+# All rights reserved.
+
+# This source code is licensed under the license found in the
+# LICENSE file in the root directory of this source tree.
+# --------------------------------------------------------
+# References:
+# ELECTRA https://github.com/google-research/electra
+# BEiT: https://github.com/microsoft/unilm/tree/master/beit
+# --------------------------------------------------------
+
+import json
+
+
+def param_groups_lrd(model, weight_decay=0.05, no_weight_decay_list=[], layer_decay=.75):
+    """
+    Parameter groups for layer-wise lr decay
+    Following BEiT: https://github.com/microsoft/unilm/blob/master/beit/optim_factory.py#L58
+    """
+    param_group_names = {}
+    param_groups = {}
+
+    num_layers = len(model.blocks) + 1
+
+    layer_scales = list(layer_decay ** (num_layers - i) for i in range(num_layers + 1))
+
+    for n, p in model.named_parameters():
+        if not p.requires_grad:
+            continue
+
+        # no decay: all 1D parameters and model specific ones
+        if p.ndim == 1 or n in no_weight_decay_list:
+            g_decay = "no_decay"
+            this_decay = 0.
+        else:
+            g_decay = "decay"
+            this_decay = weight_decay
+            
+        layer_id = get_layer_id_for_vit(n, num_layers)
+        group_name = "layer_%d_%s" % (layer_id, g_decay)
+
+        if group_name not in param_group_names:
+            this_scale = layer_scales[layer_id]
+
+            param_group_names[group_name] = {
+                "lr_scale": this_scale,
+                "weight_decay": this_decay,
+                "params": [],
+            }
+            param_groups[group_name] = {
+                "lr_scale": this_scale,
+                "weight_decay": this_decay,
+                "params": [],
+            }
+
+        param_group_names[group_name]["params"].append(n)
+        param_groups[group_name]["params"].append(p)
+
+    # print("parameter groups: \n%s" % json.dumps(param_group_names, indent=2))
+
+    return list(param_groups.values())
+
+
+def get_layer_id_for_vit(name, num_layers):
+    """
+    Assign a parameter with its layer id
+    Following BEiT: https://github.com/microsoft/unilm/blob/master/beit/optim_factory.py#L33
+    """
+    if name in ['cls_token', 'pos_embed']:
+        return 0
+    elif name.startswith('patch_embed'):
+        return 0
+    elif name.startswith('blocks'):
+        return int(name.split('.')[1]) + 1
+    else:
+        return num_layers
\ No newline at end of file

PyTorch/NLP/Vision_Transformer/util/lr_sched.py

+# Copyright (c) Meta Platforms, Inc. and affiliates.
+# All rights reserved.
+
+# This source code is licensed under the license found in the
+# LICENSE file in the root directory of this source tree.
+
+import math
+
+def adjust_learning_rate(optimizer, epoch, args):
+    """Decay the learning rate with half-cycle cosine after warmup"""
+    if epoch < args.warmup_epochs:
+        lr = args.lr * epoch / args.warmup_epochs 
+    else:
+        lr = args.min_lr + (args.lr - args.min_lr) * 0.5 * \
+            (1. + math.cos(math.pi * (epoch - args.warmup_epochs) / (args.epochs - args.warmup_epochs)))
+    for param_group in optimizer.param_groups:
+        if "lr_scale" in param_group:
+            param_group["lr"] = lr * param_group["lr_scale"]
+        else:
+            param_group["lr"] = lr
+    return lr

PyTorch/NLP/Vision_Transformer/util/misc.py

+# Copyright (c) Meta Platforms, Inc. and affiliates.
+# All rights reserved.
+
+# This source code is licensed under the license found in the
+# LICENSE file in the root directory of this source tree.
+# --------------------------------------------------------
+# References:
+# DeiT: https://github.com/facebookresearch/deit
+# BEiT: https://github.com/microsoft/unilm/tree/master/beit
+# --------------------------------------------------------
+
+import builtins
+import datetime
+import os
+import time
+from collections import defaultdict, deque
+from pathlib import Path
+
+import torch
+import torch.distributed as dist
+from torch._six import inf
+
+
+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)))
+
+
+def setup_for_distributed(is_master):
+    """
+    This function disables printing when not in master process
+    """
+    builtin_print = builtins.print
+
+    def print(*args, **kwargs):
+        force = kwargs.pop('force', False)
+        force = force or (get_world_size() > 8)
+        if is_master or force:
+            now = datetime.datetime.now().time()
+            builtin_print('[{}] '.format(now), end='')  # print with time stamp
+            builtin_print(*args, **kwargs)
+
+    builtins.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)
+
+
+def init_distributed_mode(args):
+    if args.dist_on_itp:
+        args.rank = int(os.environ['OMPI_COMM_WORLD_RANK'])
+        args.world_size = int(os.environ['OMPI_COMM_WORLD_SIZE'])
+        args.gpu = int(os.environ['OMPI_COMM_WORLD_LOCAL_RANK'])
+        args.dist_url = "tcp://%s:%s" % (os.environ['MASTER_ADDR'], os.environ['MASTER_PORT'])
+        os.environ['LOCAL_RANK'] = str(args.gpu)
+        os.environ['RANK'] = str(args.rank)
+        os.environ['WORLD_SIZE'] = str(args.world_size)
+        # ["RANK", "WORLD_SIZE", "MASTER_ADDR", "MASTER_PORT", "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'])
+    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')
+        setup_for_distributed(is_master=True)  # hack
+        args.distributed = False
+        return
+
+    args.distributed = True
+
+    torch.cuda.set_device(args.gpu)
+    args.dist_backend = 'nccl'
+    print('| distributed init (rank {}): {}, gpu {}'.format(
+        args.rank, args.dist_url, args.gpu), 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)
+
+
+class NativeScalerWithGradNormCount:
+    state_dict_key = "amp_scaler"
+
+    def __init__(self):
+        self._scaler = torch.cuda.amp.GradScaler()
+
+    def __call__(self, loss, optimizer, clip_grad=None, parameters=None, create_graph=False, update_grad=True):
+        self._scaler.scale(loss).backward(create_graph=create_graph)
+        if update_grad:
+            if clip_grad is not None:
+                assert parameters is not None
+                self._scaler.unscale_(optimizer)  # unscale the gradients of optimizer's assigned params in-place
+                norm = torch.nn.utils.clip_grad_norm_(parameters, clip_grad)
+            else:
+                self._scaler.unscale_(optimizer)
+                norm = get_grad_norm_(parameters)
+            self._scaler.step(optimizer)
+            self._scaler.update()
+        else:
+            norm = None
+        return norm
+
+    def state_dict(self):
+        return self._scaler.state_dict()
+
+    def load_state_dict(self, state_dict):
+        self._scaler.load_state_dict(state_dict)
+
+
+def get_grad_norm_(parameters, norm_type: float = 2.0) -> torch.Tensor:
+    if isinstance(parameters, torch.Tensor):
+        parameters = [parameters]
+    parameters = [p for p in parameters if p.grad is not None]
+    norm_type = float(norm_type)
+    if len(parameters) == 0:
+        return torch.tensor(0.)
+    device = parameters[0].grad.device
+    if norm_type == inf:
+        total_norm = max(p.grad.detach().abs().max().to(device) for p in parameters)
+    else:
+        total_norm = torch.norm(torch.stack([torch.norm(p.grad.detach(), norm_type).to(device) for p in parameters]), norm_type)
+    return total_norm
+
+
+def save_model(args, epoch, model, model_without_ddp, optimizer, loss_scaler):
+    output_dir = Path(args.output_dir)
+    epoch_name = str(epoch)
+    if loss_scaler is not None:
+        checkpoint_paths = [output_dir / ('checkpoint-%s.pth' % epoch_name)]
+        for checkpoint_path in checkpoint_paths:
+            to_save = {
+                'model': model_without_ddp.state_dict(),
+                'optimizer': optimizer.state_dict(),
+                'epoch': epoch,
+                'scaler': loss_scaler.state_dict(),
+                'args': args,
+            }
+
+            save_on_master(to_save, checkpoint_path)
+    else:
+        client_state = {'epoch': epoch}
+        model.save_checkpoint(save_dir=args.output_dir, tag="checkpoint-%s" % epoch_name, client_state=client_state)
+
+
+def load_model(args, model_without_ddp, optimizer, loss_scaler):
+    if args.resume:
+        if args.resume.startswith('https'):
+            checkpoint = torch.hub.load_state_dict_from_url(
+                args.resume, map_location='cpu', check_hash=True)
+        else:
+            checkpoint = torch.load(args.resume, map_location='cpu')
+        model_without_ddp.load_state_dict(checkpoint['model'])
+        print("Resume checkpoint %s" % args.resume)
+        if 'optimizer' in checkpoint and 'epoch' in checkpoint and not (hasattr(args, 'eval') and args.eval):
+            optimizer.load_state_dict(checkpoint['optimizer'])
+            args.start_epoch = checkpoint['epoch'] + 1
+            if 'scaler' in checkpoint:
+                loss_scaler.load_state_dict(checkpoint['scaler'])
+            print("With optim & sched!")
+
+
+def all_reduce_mean(x):
+    world_size = get_world_size()
+    if world_size > 1:
+        x_reduce = torch.tensor(x).cuda()
+        dist.all_reduce(x_reduce)
+        x_reduce /= world_size
+        return x_reduce.item()
+    else:
+        return x
\ No newline at end of file

PyTorch/NLP/Vision_Transformer/util/pos_embed.py

+# Copyright (c) Meta Platforms, Inc. and affiliates.
+# All rights reserved.
+
+# This source code is licensed under the license found in the
+# LICENSE file in the root directory of this source tree.
+# --------------------------------------------------------
+# Position embedding utils
+# --------------------------------------------------------
+
+import numpy as np
+
+import torch
+
+# --------------------------------------------------------
+# 2D sine-cosine position embedding
+# References:
+# Transformer: https://github.com/tensorflow/models/blob/master/official/nlp/transformer/model_utils.py
+# MoCo v3: https://github.com/facebookresearch/moco-v3
+# --------------------------------------------------------
+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