finetune.py

#!/usr/bin/env python3
""" ImageNet Training Script

This is intended to be a lean and easily modifiable ImageNet training script that reproduces ImageNet
training results with some of the latest networks and training techniques. It favours canonical PyTorch
and standard Python style over trying to be able to 'do it all.' That said, it offers quite a few speed
and training result improvements over the usual PyTorch example scripts. Repurpose as you see fit.

This script was started from an early version of the PyTorch ImageNet example
(https://github.com/pytorch/examples/tree/master/imagenet)

NVIDIA CUDA specific speedups adopted from NVIDIA Apex examples
(https://github.com/NVIDIA/apex/tree/master/examples/imagenet)

Taken from https://github.com/rwightman
Copyright 2020 Ross Wightman (https://github.com/rwightman)
"""
import argparse
import copy
import logging
import os
import shutil
import time
import sys
from collections import OrderedDict
from contextlib import suppress
from datetime import datetime
from functools import partial

import torch
import torch.nn as nn
import torchvision.utils
import yaml
from torch.nn.parallel import DistributedDataParallel as NativeDDP

from timm import utils
from timm.data import create_dataset, create_loader, resolve_data_config, Mixup, FastCollateMixup, AugMixDataset
from timm.loss import JsdCrossEntropy, SoftTargetCrossEntropy, BinaryCrossEntropy, \
    LabelSmoothingCrossEntropy

from timm.models import create_model, safe_model_name, resume_checkpoint, load_checkpoint, \
    convert_splitbn_model, convert_sync_batchnorm, model_parameters, set_fast_norm

from timm.optim import create_optimizer_v2, optimizer_kwargs
from timm.scheduler import create_scheduler_v2, scheduler_kwargs
from timm.utils import ApexScaler, NativeScaler

try:
    from apex import amp
    from apex.parallel import DistributedDataParallel as ApexDDP
    from apex.parallel import convert_syncbn_model
    has_apex = True
except ImportError:
    has_apex = False

has_native_amp = False
try:
    if getattr(torch.cuda.amp, 'autocast') is not None:
        has_native_amp = True
except AttributeError:
    pass

try:
    import wandb
    has_wandb = True
except ImportError:
    has_wandb = False

try:
    from functorch.compile import memory_efficient_fusion
    has_functorch = True
except ImportError as e:
    has_functorch = False

#import nvidia_smi

import numpy as np
from finetune_utils.utils import extend_metrics, prepare_model_for_finetuning, get_dataset_path, \
                        get_icgen_dataset_info_json, get_number_of_classes, extend_metrics,\
                        compute_gradient_norm
from finetune_utils.finetuning_stategies import SPRegularization, BatchSpectralShrinkage, \
                                BehavioralRegularization, CoTuningLoss, Relationship, \
                                convert_to_stoch_norm

try:
    from syne_tune import Reporter
    has_synetune = True
except ImportError:
    has_synetune = False

_logger = logging.getLogger('train')

# The first arg parser parses out only the --config argument, this argument is used to
# load a yaml file containing key-values that override the defaults for the main parser below
config_parser = parser = argparse.ArgumentParser(description='Training Config', add_help=False)
parser.add_argument('-c', '--config', default='', type=str, metavar='FILE',
                    help='YAML config file specifying default arguments')

parser = argparse.ArgumentParser(description='PyTorch ImageNet Training')

# Dataset parameters
group = parser.add_argument_group('Dataset parameters')
# Keep this argument outside of the dataset group because it is positional.
parser.add_argument('data_dir', metavar='DIR', default='datasets/inat', nargs='?',
                    help='path to the core datasets (e.g. /work/dlclarge1/pineda-zap-data/tfds-icgen/core_datasets')

group.add_argument('--dataset_augmentation_path', '-da', metavar='NAME', default='',
                    help='if specified, meta-information is taken from the individual dataset augmentation path, e.g. "/work/dlclarge1/pineda-zap-data/tfds-icgen/1". If empty (default), '
                         'only datasets located under <data_dir> are considered.')

group.add_argument('--dataset', '-d', metavar='NAME', default='', help='dataset name, e.g. tfds/eurosat/rgb')

group.add_argument('--train-split', metavar='NAME', default='train',
                    help='dataset train split (default: train)')
group.add_argument('--val-split', metavar='NAME', default='validation',
                    help='dataset validation split (default: validation)')
group.add_argument('--dataset-download', action='store_true', default=False,
                    help='Allow download of dataset for torch/ and tfds/ datasets that support it.')
group.add_argument('--class-map', default='', type=str, metavar='FILENAME',
                    help='path to class to idx mapping file (default: "")')

# Model parameters
group = parser.add_argument_group('Model parameters')
group.add_argument('--model', default='resnet50', type=str, metavar='MODEL',
                    help='Name of model to train (default: "resnet50")')
group.add_argument('--pretrained', action='store_true', default=True,
                    help='Start with pretrained version of specified network (if avail)')
group.add_argument('--initial-checkpoint', default='', type=str, metavar='PATH',
                    help='Initialize model from this checkpoint (default: none)')
group.add_argument('--resume', default='', type=str, metavar='PATH',
                    help='Resume full model and optimizer state from checkpoint (default: none)')
group.add_argument('--no-resume-opt', action='store_true', default=False,
                    help='prevent resume of optimizer state when resuming model')
group.add_argument('--num-classes', '--num_classes', type=int, default=None, metavar='N',
                    help='number of label classes (Model default if None)')
group.add_argument('--gp', default=None, type=str, metavar='POOL',
                    help='Global pool type, one of (fast, avg, max, avgmax, avgmaxc). Model default if None.')
group.add_argument('--img-size', '--img_size', type=int, default=None, metavar='N',
                    help='Image size (default: None => model default)')
group.add_argument('--in-chans', '--in_chans', type=int, default=None, metavar='N',
                    help='Image input channels (default: None => 3)')
group.add_argument('--input-size', '--input_size', default=None, nargs=3, type=int,
                    metavar='N N N', help='Input all image dimensions (d h w, e.g. --input-size 3 224 224), uses model default if empty')
group.add_argument('--crop-pct', default=None, type=float,
                    metavar='N', help='Input image center crop percent (for validation only)')
group.add_argument('--mean', type=float, nargs='+', default=None, metavar='MEAN',
                    help='Override mean pixel value of dataset')
group.add_argument('--std', type=float, nargs='+', default=None, metavar='STD',
                    help='Override std deviation of dataset')
group.add_argument('--interpolation', default='', type=str, metavar='NAME',
                    help='Image resize interpolation type (overrides model)')
group.add_argument('-b', '--batch-size', '--batch_size', type=int, default=128, metavar='N',
                    help='Input batch size for training (default: 128)')
group.add_argument('-vb', '--validation-batch-size', '--validation_batch_size', type=int, default=None, metavar='N',
                    help='Validation batch size override (default: None)')
group.add_argument('--channels-last', action='store_true', default=False,
                    help='Use channels_last memory layout')
scripting_group = group.add_mutually_exclusive_group()
scripting_group.add_argument('--torchscript', dest='torchscript', action='store_true',
                    help='torch.jit.script the full model')
scripting_group.add_argument('--aot-autograd', default=False, action='store_true',
                    help="Enable AOT Autograd support. (It's recommended to use this option with `--fuser nvfuser` together)")
group.add_argument('--fuser', default='', type=str,
                    help="Select jit fuser. One of ('', 'te', 'old', 'nvfuser')")
group.add_argument('--fast-norm', default=False, action='store_true',
                    help='enable experimental fast-norm')
group.add_argument('--grad-checkpointing', action='store_true', default=False,
                    help='Enable gradient checkpointing through model blocks/stages')

# Optimizer parameters
group = parser.add_argument_group('Optimizer parameters')
group.add_argument('--opt', default='sgd', type=str, metavar='OPTIMIZER',
                    help='Optimizer (default: "sgd")')
group.add_argument('--opt-eps', '--opt_eps', default=None, type=float, metavar='EPSILON',
                    help='Optimizer Epsilon (default: None, use opt default)')
group.add_argument('--opt-betas', '--opt_betas', default=None, type=float, nargs='+', metavar='BETA',
                    help='Optimizer Betas (default: None, use opt default)')
group.add_argument('--momentum', type=float, default=0.9, metavar='M',
                    help='Optimizer momentum (default: 0.9)')
group.add_argument('--weight-decay', '--weight_decay', type=float, default=2e-5,
                    help='weight decay (default: 2e-5)')
group.add_argument('--clip-grad', '--clip_grad', type=float, default=None, metavar='NORM',
                    help='Clip gradient norm (default: None, no clipping)')
group.add_argument('--clip-mode', '--clip_mode', type=str, default='norm',
                    help='Gradient clipping mode. One of ("norm", "value", "agc")')
group.add_argument('--layer-decay', '--layer_decay', type=float, default=None,
                    help='layer-wise learning rate decay (default: None)')

# Learning rate schedule parameters
group = parser.add_argument_group('Learning rate schedule parameters')
group.add_argument('--sched', type=str, default='cosine', metavar='SCHEDULER',
                    help='LR scheduler (default: "step"')
group.add_argument('--sched-on-updates', action='store_true', default=False,
                    help='Apply LR scheduler step on update instead of epoch end.')
group.add_argument('--lr', type=float, default=None, metavar='LR',
                    help='learning rate, overrides lr-base if set (default: None)')
group.add_argument('--lr-base', '--lr_base',type=float, default=0.1, metavar='LR',
                    help='base learning rate: lr = lr_base * global_batch_size / base_size')
group.add_argument('--lr-base-size', '--lr_base_size', type=int, default=256, metavar='DIV',
                    help='base learning rate batch size (divisor, default: 256).')
group.add_argument('--lr-base-scale', '--lr_base_scale', type=str, default='', metavar='SCALE',
                    help='base learning rate vs batch_size scaling ("linear", "sqrt", based on opt if empty)')
group.add_argument('--lr-noise', type=float, nargs='+', default=None, metavar='pct, pct',
                    help='learning rate noise on/off epoch percentages')
group.add_argument('--lr-noise-pct', type=float, default=0.67, metavar='PERCENT',
                    help='learning rate noise limit percent (default: 0.67)')
group.add_argument('--lr-noise-std', type=float, default=1.0, metavar='STDDEV',
                    help='learning rate noise std-src (default: 1.0)')
group.add_argument('--lr-cycle-mul', type=float, default=1.0, metavar='MULT',
                    help='learning rate cycle len multiplier (default: 1.0)')
group.add_argument('--lr-cycle-decay', type=float, default=0.5, metavar='MULT',
                    help='amount to decay each learning rate cycle (default: 0.5)')
group.add_argument('--lr-cycle-limit', type=int, default=1, metavar='N',
                    help='learning rate cycle limit, cycles enabled if > 1')
group.add_argument('--lr-k-decay', '--lr_k_decay', type=float, default=1.0,
                    help='learning rate k-decay for cosine/poly (default: 1.0)')
group.add_argument('--warmup-lr', '--warmup_lr' ,type=float, default=1e-5, metavar='LR',
                    help='warmup learning rate (default: 1e-5)')
group.add_argument('--min-lr', '--min_lr', type=float, default=0, metavar='LR',
                    help='lower lr bound for cyclic schedulers that hit 0 (default: 0)')
group.add_argument('--epochs', type=int, default=300, metavar='N',
                    help='number of epochs to train (default: 300)')
group.add_argument('--epoch-repeats', type=float, default=0., metavar='N',
                    help='epoch repeat multiplier (number of times to repeat dataset epoch per train epoch).')
group.add_argument('--start-epoch', default=None, type=int, metavar='N',
                    help='manual epoch number (useful on restarts)')
group.add_argument('--decay-milestones', default=[90, 180, 270], type=int, nargs='+', metavar="MILESTONES",
                    help='list of decay epoch indices for multistep lr. must be increasing')
group.add_argument('--decay-epochs', '--decay_epochs', type=float, default=90, metavar='N',
                    help='epoch interval to decay LR')
group.add_argument('--warmup-epochs', '--warmup_epochs', type=int, default=5, metavar='N',
                    help='epochs to warmup LR, if scheduler supports')
group.add_argument('--warmup-prefix', action='store_true', default=False,
                    help='Exclude warmup period from decay schedule.'),
group.add_argument('--cooldown-epochs', type=int, default=0, metavar='N',
                    help='epochs to cooldown LR at min_lr, after cyclic schedule ends')
group.add_argument('--patience-epochs', '--patience_epochs', type=int, default=10, metavar='N',
                    help='patience epochs for Plateau LR scheduler (default: 10)')
group.add_argument('--decay-rate', '--dr', '--decay_rate', type=float, default=0.1, metavar='RATE',
                    help='LR decay rate (default: 0.1)')

# Augmentation & regularization parameters
group = parser.add_argument_group('Augmentation and regularization parameters')
group.add_argument('--no-aug', action='store_true', default=False,
                    help='Disable all training augmentation, override other train aug args')
group.add_argument('--scale', type=float, nargs='+', default=[0.08, 1.0], metavar='PCT',
                    help='Random resize scale (default: 0.08 1.0)')
group.add_argument('--ratio', type=float, nargs='+', default=[3./4., 4./3.], metavar='RATIO',
                    help='Random resize aspect ratio (default: 0.75 1.33)')
group.add_argument('--hflip', type=float, default=0.5,
                    help='Horizontal flip training aug probability')
group.add_argument('--vflip', type=float, default=0.,
                    help='Vertical flip training aug probability')
group.add_argument('--color-jitter', type=float, default=0.4, metavar='PCT',
                    help='Color jitter factor (default: 0.4)')


group.add_argument( '--auto_augment', type=str, default=None, metavar='NAME',
                    help='Use AutoAugment policy. "v0" or "original". (default: None)'),
group.add_argument('--trivial_augment', action='store_true', default=False,
                    help='Use TrivialAgument". (default: False)'),


# TODO: implement
group.add_argument('--random_augment', action='store_true', default=False,
                    help='Use RandAugmentation policy. (default: None)')
group.add_argument('--ra_num_ops', type=int, default=2,
                    help='todo. (default: None)')
group.add_argument('--ra_magnitude', type=int, default=8,
                    help='todo. (default: None)')
group.add_argument('--ra_num_magnitude_bins', type=int, default=31,
                    help='todo. (default: None)')
group.add_argument('--ra_interpolation', type=str, default='nearest',
                    help='todo. (default: None)')

group.add_argument('--aug-repeats', type=float, default=0,
                    help='Number of augmentation repetitions (distributed training only) (default: 0)')
group.add_argument('--aug-splits', type=int, default=0,
                    help='Number of augmentation splits (default: 0, valid: 0 or >=2)')
group.add_argument('--jsd-loss', action='store_true', default=False,
                    help='Enable Jensen-Shannon Divergence + CE loss. Use with `--aug-splits`.')
group.add_argument('--bce-loss', action='store_true', default=False,
                    help='Enable BCE loss w/ Mixup/CutMix use.')
group.add_argument('--bce-target-thresh', type=float, default=None,
                    help='Threshold for binarizing softened BCE targets (default: None, disabled)')
group.add_argument('--reprob', type=float, default=0., metavar='PCT',
                    help='Random erase prob (default: 0.)')
group.add_argument('--remode', type=str, default='pixel',
                    help='Random erase mode (default: "pixel")')
group.add_argument('--recount', type=int, default=1,
                    help='Random erase count (default: 1)')
group.add_argument('--resplit', action='store_true', default=False,
                    help='Do not random erase first (clean) augmentation split')
group.add_argument('--mixup', type=float, default=0.0,
                    help='mixup alpha, mixup enabled if > 0. (default: 0.)')
group.add_argument('--cutmix', type=float, default=0.0,
                    help='cutmix alpha, cutmix enabled if > 0. (default: 0.)')
group.add_argument('--cutmix-minmax', type=float, nargs='+', default=None,
                    help='cutmix min/max ratio, overrides alpha and enables cutmix if set (default: None)')
group.add_argument('--mixup-prob', '--mixup_prob', type=float, default=1.0,
                    help='Probability of performing mixup or cutmix when either/both is enabled')
group.add_argument('--mixup-switch-prob', type=float, default=0.5,
                    help='Probability of switching to cutmix when both mixup and cutmix enabled')
group.add_argument('--mixup-mode', type=str, default='batch',
                    help='How to apply mixup/cutmix params. Per "batch", "pair", or "elem"')
group.add_argument('--mixup-off-epoch', default=0, type=int, metavar='N',
                    help='Turn off mixup after this epoch, disabled if 0 (default: 0)')
group.add_argument('--smoothing', type=float, default=0.1,
                    help='Label smoothing (default: 0.1)')
group.add_argument('--train-interpolation', type=str, default='random',
                    help='Training interpolation (random, bilinear, bicubic default: "random")')
group.add_argument('--drop', type=float, default=0.0, metavar='PCT',
                    help='Dropout rate (default: 0.)')
group.add_argument('--drop-connect', type=float, default=None, metavar='PCT',
                    help='Drop connect rate, DEPRECATED, use drop-path (default: None)')
group.add_argument('--drop-path', '--drop_path', type=float, default=None, metavar='PCT',
                    help='Drop path rate (default: None)')
group.add_argument('--drop-block', type=float, default=None, metavar='PCT',
                    help='Drop block rate (default: None)')

# Batch norm parameters (only works with gen_efficientnet based models currently)
group = parser.add_argument_group('Batch norm parameters', 'Only works with gen_efficientnet based models currently.')
group.add_argument('--bn-momentum', type=float, default=None,
                    help='BatchNorm momentum override (if not None)')
group.add_argument('--bn-eps', type=float, default=None,
                    help='BatchNorm epsilon override (if not None)')
group.add_argument('--sync-bn', action='store_true',
                    help='Enable NVIDIA Apex or Torch synchronized BatchNorm.')
group.add_argument('--dist-bn', type=str, default='reduce',
                    help='Distribute BatchNorm stats between nodes after each epoch ("broadcast", "reduce", or "")')
group.add_argument('--split-bn', action='store_true',
                    help='Enable separate BN layers per augmentation split.')

# Model Exponential Moving Average
group = parser.add_argument_group('Model exponential moving average parameters')
group.add_argument('--model-ema', '--model_ema', action='store_true', default=False,
                    help='Enable tracking moving average of model weights')
group.add_argument('--model-ema-force-cpu', action='store_true', default=False,
                    help='Force ema to be tracked on CPU, rank=0 node only. Disables EMA validation.')
group.add_argument('--model-ema-decay', type=float, default=0.9998,
                    help='decay factor for model weights moving average (default: 0.9998)')

# Misc
group = parser.add_argument_group('Miscellaneous parameters')
group.add_argument('--pct_to_freeze', type=float, default=1.0,
                   help='percentage of layers to freeze')
group.add_argument('--seed', type=int, default=42, metavar='S',
                    help='random seed (default: 42)')
group.add_argument('--worker-seeding', type=str, default='all',
                    help='worker seed mode (default: all)')
group.add_argument('--log-interval', '--log_interval', type=int, default=50, metavar='N',
                    help='how many batches to wait before logging training status')
group.add_argument('--recovery-interval', type=int, default=0, metavar='N',
                    help='how many batches to wait before writing recovery checkpoint')
group.add_argument('--checkpoint-hist', '--checkpoint_hist', type=int, default=10, metavar='N',
                    help='number of checkpoints to keep (default: 10)')
group.add_argument('-j', '--workers', type=int, default=4, metavar='N',
                    help='how many training processes to use (default: 4)')
group.add_argument('--save-images', action='store_true', default=False,
                    help='save images of input bathes every log interval for debugging')
group.add_argument('--amp', action='store_true', default=False,
                    help='use NVIDIA Apex AMP or Native AMP for mixed precision training')
group.add_argument('--amp-dtype', default='float16', type=str,
                    help='lower precision AMP dtype (default: float16)')
group.add_argument('--amp-impl', default='native', type=str,
                    help='AMP impl to use, "native" or "apex" (default: native)')
group.add_argument('--no-ddp-bb', action='store_true', default=False,
                    help='Force broadcast buffers for native DDP to off.')
group.add_argument('--pin-mem', action='store_true', default=False,
                    help='Pin CPU memory in DataLoader for more efficient (sometimes) transfer to GPU.')
group.add_argument('--no-prefetcher', action='store_true', default=False,
                    help='disable fast prefetcher')
group.add_argument('--output', default='', type=str, metavar='PATH',
                    help='path to output folder (default: none, current dir)')
group.add_argument('--experiment', default='', type=str, metavar='NAME',
                    help='name of train experiment, name of sub-folder for output')
group.add_argument('--eval-metric', default='top1', type=str, metavar='EVAL_METRIC',
                    help='Best metric (default: "top1"')
group.add_argument('--tta', type=int, default=0, metavar='N',
                    help='Test/inference time augmentation (oversampling) factor. 0=None (default: 0)')
group.add_argument("--local_rank", default=0, type=int)
group.add_argument('--use-multi-epochs-loader', action='store_true', default=False,
                    help='use the multi-epochs-loader to save time at the beginning of every epoch')
group.add_argument('--log-wandb', action='store_true', default=False,
                    help='log training and validation metrics to wandb')
group.add_argument('--test_mode', action='store_true', default=False,
                    help='this activates the test mode which is more efficient than normal mode by deactivating checkpointing and setting'\
                         'epochs=1 and batches per epoch=1')
group.add_argument('--persistent_workers', action='store_true', default=False,
                    help='whether to use persistent workers for the dataloader')
group.add_argument('--project_name', type=str, default=False,
                    help='Project name for wandb')
group.add_argument('--epochs_step', type=int, default=-1,
                    help='how many actual epochs to run, -1 means all epochs as defined by --epochs')
group.add_argument('--initial_test', action='store_true', default=False)
group.add_argument('--report_synetune', type=int, default=0)
group.add_argument('--st_checkpoint_dir', type=str, default=None, help='path to the checkpoint directory of the model'
                                                                       'when using synetune')

# finetuning strategies
group.add_argument('--linear_probing', action='store_true', default=False,
                   help='whether to apply linear probe finetuning strategy')
group.add_argument('--stoch_norm', action='store_true', default=False,
                   help='whether to apply stochastic norm')
group.add_argument('--sp_reg', type=float, default=0.0,
                    help='SP regularization. Default = 0.')
group.add_argument('--cotuning_reg', type=float, default=0.0,
                    help='Co-Tuning regularization. Default = 0.')
group.add_argument('--bss_reg', type=float, default=0.0,
                    help='BSS regularization. Default = 0.')
group.add_argument('--delta_reg', type=float, default=0.0,
                    help='DELTA regularization. Default = 0.')

current_dir = os.path.dirname(os.path.realpath(__file__))
global_epoch = 0
#if has_synetune:
#    report = Reporter()
#else:
#    report = None

def _parse_args():
    # Do we have a config file to parse?
    args_config, remaining = config_parser.parse_known_args()
    if args_config.config:
        with open(args_config.config, 'r') as f:
            cfg = yaml.safe_load(f)
            parser.set_defaults(**cfg)

    # The main arg parser parses the rest of the args, the usual
    # defaults will have been overridden if config file specified.
    args = parser.parse_args(remaining)

    # Cache the args as a text string to save them in the output dir later
    args_text = yaml.safe_dump(args.__dict__, default_flow_style=False)
    return args, args_text


def main(args, args_text):


    utils.setup_default_logging()
    device_count = torch.cuda.device_count()
    if torch.cuda.is_available():
        torch.backends.cuda.matmul.allow_tf32 = True
        torch.backends.cudnn.benchmark = True
    args.prefetcher = not args.no_prefetcher
    device = utils.init_distributed_device(args)
    if args.distributed:
        _logger.info(
            'Training in distributed mode with multiple processes, 1 device per process.'
            f'Process {args.rank}, total {args.world_size}, device {args.device}.')
    else:
        _logger.info(f'Training with a single process on 1 device ({args.device}).')
    assert args.rank >= 0

    if utils.is_primary(args) and args.log_wandb:
        if has_wandb:
            project_name = args.project_name
            wandb.init(project=project_name, name=args.experiment, config=args)
        else:
            _logger.warning(
                "You've requested to log metrics to wandb but package not found. "
                "Metrics not being logged to wandb, try `pip install wandb`")

    # resolve AMP arguments based on PyTorch / Apex availability
    use_amp = None
    amp_dtype = torch.float16
    if args.amp:
        if args.amp_impl == 'apex':
            assert has_apex, 'AMP impl specified as APEX but APEX is not installed.'
            use_amp = 'apex'
            assert args.amp_dtype == 'float16'
        else:
            assert has_native_amp, 'Please update PyTorch to a version with native AMP (or use APEX).'
            use_amp = 'native'
            assert args.amp_dtype in ('float16', 'bfloat16')
        if args.amp_dtype == 'bfloat16':
            amp_dtype = torch.bfloat16

    utils.random_seed(args.seed, args.rank)
    
    if args.test_mode:
        print("------ Test Mode active ------")
        args.epochs = 1

    if has_synetune and args.report_synetune:
        global report
        report = Reporter()
        args.data_dir = os.path.join(current_dir, args.data_dir)
        print(args.data_dir)

    if args.fuser:
        utils.set_jit_fuser(args.fuser)
    if args.fast_norm:
        set_fast_norm()

    in_chans = 3
    if args.in_chans is not None:
        in_chans = args.in_chans
    elif args.input_size is not None:
        in_chans = args.input_size[0]

    if args.dataset_augmentation_path:
        try:
            dataset_aug_path = get_dataset_path(args.dataset_augmentation_path, args.dataset)
            # dataset_path = get_dataset_path(args.data_dir, args.dataset)
            icgen_dataset_info = get_icgen_dataset_info_json(dataset_aug_path, args.dataset)
    
            actual_num_classes = icgen_dataset_info["number_classes"]
            original_num_classes = get_number_of_classes(dataset_aug_path)
            args.num_classes = actual_num_classes
            extra_dataset_info = {"icgen_dataset_info": icgen_dataset_info}

            assert args.dataset.startswith("tfdsicgn"), "dataset_augmentation_path is only supported for tfds_icgn datasets"
    
            print(f"Number of classes in original dataset: {original_num_classes}")
            print(f"Number of classes sampled from: {actual_num_classes}")
            print(f"Number of train samples: {icgen_dataset_info['number_train_samples_per_class']*actual_num_classes}")
            print(f"Number of test samples: {icgen_dataset_info['number_test_samples_per_class']*actual_num_classes}")
    
        except Exception as e:
            raise ValueError(f"Error reading dataset information: {e}")
    else:
        assert args.num_classes is not None, "num_classes not inferred from data, please set it manually"
        # we don't need sub-sampling the datasets (only used when ICGen augmentations are used and dataset_augmentation_path is set)
        extra_dataset_info = {}

    model = create_model(
        args.model,
        pretrained=args.pretrained,
        in_chans=in_chans,
        num_classes=args.num_classes,
        drop_rate=args.drop,
        drop_path_rate=args.drop_path,
        drop_block_rate=args.drop_block,
        global_pool=args.gp,
        bn_momentum=args.bn_momentum,
        bn_eps=args.bn_eps,
        scriptable=args.torchscript,
        checkpoint_path=args.initial_checkpoint,
    )

    if args.grad_checkpointing:
        model.set_grad_checkpointing(enable=True)

    if utils.is_primary(args):
        _logger.info(
            f'Model {safe_model_name(args.model)} created, param count:{sum([m.numel() for m in model.parameters()])}')

    data_config = resolve_data_config(vars(args), model=model, verbose=utils.is_primary(args))

    #correct data config mean
    if args.in_chans == 1:
        data_config["mean"] = data_config["mean"][:1]
        data_config["std"] = data_config["std"][:1]

    # setup augmentation batch splits for contrastive loss or split bn
    num_aug_splits = 0
    if args.aug_splits > 0:
        assert args.aug_splits > 1, 'A split of 1 makes no sense'
        num_aug_splits = args.aug_splits

    # enable split bn (separate bn stats per batch-portion)
    if args.split_bn:
        assert num_aug_splits > 1 or args.resplit
        model = convert_splitbn_model(model, max(num_aug_splits, 2))

    # move model to GPU, enable channels last layout if set
    model.to(device=device)
    if args.channels_last:
        model.to(memory_format=torch.channels_last)

    # setup synchronized BatchNorm for distributed training
    if args.distributed and args.sync_bn:
        args.dist_bn = ''  # disable dist_bn when sync BN active
        assert not args.split_bn
        if has_apex and use_amp == 'apex':
            # Apex SyncBN used with Apex AMP
            # WARNING this won't currently work with models using BatchNormAct2d
            model = convert_syncbn_model(model)
        else:
            model = convert_sync_batchnorm(model)
        if utils.is_primary(args):
            _logger.info(
                'Converted model to use Synchronized BatchNorm. WARNING: You may have issues if using '
                'zero initialized BN layers (enabled by default for ResNets) while sync-bn enabled.')

    if args.torchscript:
        assert not use_amp == 'apex', 'Cannot use APEX AMP with torchscripted model'
        assert not args.sync_bn, 'Cannot use SyncBatchNorm with torchscripted model'
        model = torch.jit.script(model)

    if args.aot_autograd:
        assert has_functorch, "functorch is needed for --aot-autograd"
        model = memory_efficient_fusion(model)

    if args.lr is None:
        global_batch_size = args.batch_size * args.world_size
        batch_ratio = global_batch_size / args.lr_base_size
        if not args.lr_base_scale:
            on = args.opt.lower()
            args.lr_base_scale = 'sqrt' if any([o in on for o in ('ada', 'lamb')]) else 'linear'
        if args.lr_base_scale == 'sqrt':
            batch_ratio = batch_ratio ** 0.5
        args.lr = args.lr_base * batch_ratio
        if utils.is_primary(args):
            _logger.info(
                f'Learning rate ({args.lr}) calculated from base learning rate ({args.lr_base}) '
                f'and global batch size ({global_batch_size}) with {args.lr_base_scale} scaling.')

    if args.linear_probing:
        pct_to_freeze = 1.0
    else:
        pct_to_freeze = args.pct_to_freeze

    return_features =  False
    return_source_output = False

    if args.bss_reg or args.delta_reg or args.cotuning_reg:
        return_features = True
    if args.delta_reg or args.cotuning_reg:
        return_source_output = True

    #to do: add the regularization weight to all methods
    if args.stoch_norm:
        model = convert_to_stoch_norm(model)
        
    model, head_name = prepare_model_for_finetuning(model,
                                                    num_classes=args.num_classes,
                                                    pct_to_freeze=pct_to_freeze,
                                                    return_features=return_features,
                                                    return_source_output=return_source_output
                                                    )

    backbone_regularizations = []
    if args.sp_reg > 0.0:
        backbone_regularizations.append(SPRegularization(model, head_name, args.sp_reg))
    if args.bss_reg > 0.0:
        backbone_regularizations.append(BatchSpectralShrinkage(regularization_weight=args.bss_reg))
    if args.delta_reg > 0.0:
        source_model = copy.deepcopy(model)
        backbone_regularizations.append(BehavioralRegularization(source_model, regularization_weight=args.delta_reg))
    if args.cotuning_reg > 0.0:
        compute_relationship = True
        source_model = copy.deepcopy(model)
        backbone_regularizations.append(CoTuningLoss(args.cotuning_reg))
    else:
        compute_relationship = False

    optimizer = create_optimizer_v2(model, **optimizer_kwargs(cfg=args))

    # setup automatic mixed-precision (AMP) loss scaling and op casting
    amp_autocast = suppress  # do nothing
    loss_scaler = None
    if use_amp == 'apex':
        assert device.type == 'cuda'
        model, optimizer = amp.initialize(model, optimizer, opt_level='O1')
        loss_scaler = ApexScaler()
        if utils.is_primary(args):
            _logger.info('Using NVIDIA APEX AMP. Training in mixed precision.')
    elif use_amp == 'native':
        amp_autocast = partial(torch.autocast, device_type=device.type, dtype=amp_dtype)
        if device.type == 'cuda':
            loss_scaler = NativeScaler()
        if utils.is_primary(args):
            _logger.info('Using native Torch AMP. Training in mixed precision.')
    else:
        if utils.is_primary(args):
            _logger.info('AMP not enabled. Training in float32.')

    # optionally resume from a checkpoint
    resume_epoch = None
    if args.resume:
        resume_epoch = resume_checkpoint(
            model,
            args.resume,
            optimizer=None if args.no_resume_opt else optimizer,
            loss_scaler=None if args.no_resume_opt else loss_scaler,
            log_info=utils.is_primary(args),
        )

    # setup exponential moving average of model weights, SWA could be used here too
    model_ema = None
    if args.model_ema:
        # Important to create EMA model after cuda(), DP wrapper, and AMP but before DDP wrapper
        model_ema = utils.ModelEmaV2(
            model, decay=args.model_ema_decay, device='cpu' if args.model_ema_force_cpu else None)
        if args.resume:
            load_checkpoint(model_ema.module, args.resume, use_ema=True)

    # setup distributed training
    if args.distributed:
        if has_apex and use_amp == 'apex':
            # Apex DDP preferred unless native amp is activated
            if utils.is_primary(args):
                _logger.info("Using NVIDIA APEX DistributedDataParallel.")
            model = ApexDDP(model, delay_allreduce=True)
        else:
            if utils.is_primary(args):
                _logger.info("Using native Torch DistributedDataParallel.")
            model = NativeDDP(model, device_ids=[device], broadcast_buffers=not args.no_ddp_bb)
        # NOTE: EMA model does not need to be wrapped by DDP
        
    # create the train and eval datasets
    dataset_train = create_dataset(
        args.dataset,
        root=args.data_dir,
        split=args.train_split,
        is_training=True,
        class_map=args.class_map,
        download=args.dataset_download,
        batch_size=args.batch_size,
        seed=args.seed,
        repeats=args.epoch_repeats,
        input_img_mode="L" if in_chans == 1 else "RGB",
        **extra_dataset_info
    )

    dataset_eval = create_dataset(
        args.dataset,
        root=args.data_dir,
        split=args.val_split,
        is_training=False,
        class_map=args.class_map,
        download=args.dataset_download,
        batch_size=args.batch_size,
        input_img_mode="L" if in_chans == 1 else "RGB",
        **extra_dataset_info
    )
    # setup mixup / cutmix
    collate_fn = None
    mixup_fn = None
    mixup_active = args.mixup > 0 or args.cutmix > 0. or args.cutmix_minmax is not None
    if mixup_active:
        mixup_args = dict(
            mixup_alpha=args.mixup,
            cutmix_alpha=args.cutmix,
            cutmix_minmax=args.cutmix_minmax,
            prob=args.mixup_prob,
            switch_prob=args.mixup_switch_prob,
            mode=args.mixup_mode,
            label_smoothing=args.smoothing,
            num_classes=args.num_classes
        )
        if args.prefetcher:
            assert not num_aug_splits  # collate conflict (need to support deinterleaving in collate mixup)
            collate_fn = FastCollateMixup(**mixup_args)
        else:
            mixup_fn = Mixup(**mixup_args)

    # wrap dataset in AugMix helper
    if num_aug_splits > 1:
        dataset_train = AugMixDataset(dataset_train, num_splits=num_aug_splits)

    # create data loaders w/ augmentation pipeline
    train_interpolation = args.train_interpolation
    if args.no_aug or not train_interpolation:
        train_interpolation = data_config['interpolation']
    loader_train = create_loader(
        dataset_train,
        input_size=data_config['input_size'],
        in_chans=in_chans,
        batch_size=args.batch_size,
        is_training=True,
        use_prefetcher=args.prefetcher,
        no_aug=args.no_aug,
        re_prob=args.reprob,
        re_mode=args.remode,
        re_count=args.recount,
        re_split=args.resplit,
        scale=args.scale,
        ratio=args.ratio,
        hflip=args.hflip,
        vflip=args.vflip,
        color_jitter=args.color_jitter,
        auto_augment=args.auto_augment,
        num_aug_repeats=args.aug_repeats,
        num_aug_splits=num_aug_splits,
        interpolation=train_interpolation,
        mean=data_config['mean'],
        std=data_config['std'],
        num_workers=args.workers,
        distributed=args.distributed,
        collate_fn=collate_fn,
        pin_memory=args.pin_mem,
        device=device,
        use_multi_epochs_loader=args.use_multi_epochs_loader,
        worker_seeding=args.worker_seeding,
        trivial_augment=args.trivial_augment,
        random_augment=args.random_augment,
        ra_num_ops=args.ra_num_ops,
        ra_magnitude=args.ra_magnitude,
        persistent_workers=args.persistent_workers,
    )

    eval_workers = args.workers
    if args.distributed and ('tfds' in args.dataset or 'wds' in args.dataset):
        # FIXME reduces validation padding issues when using TFDS, WDS w/ workers and distributed training
        eval_workers = min(2, args.workers)
    loader_eval = create_loader(
        dataset_eval,
        input_size=data_config['input_size'],
        in_chans=in_chans,
        batch_size=args.validation_batch_size or args.batch_size,
        is_training=False,
        use_prefetcher=args.prefetcher,
        interpolation=data_config['interpolation'],
        mean=data_config['mean'],
        std=data_config['std'],
        num_workers=eval_workers,
        distributed=args.distributed,
        crop_pct=data_config['crop_pct'],
        pin_memory=args.pin_mem,
        device=device,
        persistent_workers=args.persistent_workers,
    )

    # relationship for the co-tuning loss
    if compute_relationship:
        file_name = os.path.split(args.data_dir)[1] + "_" \
                        + args.model + "_" \
                        + args.dataset_augmentation_path.replace("/","_") + "_" \
                        + args.dataset.replace("/","_") + ".npy"
        output_dir = utils.get_outdir(os.path.join(args.output, "..", "relationships") if args.output else './output/relationships')
        relationship = Relationship(loader_train, source_model, device, cache = os.path.join(output_dir, file_name))
    else:
        relationship = None
    # setup loss function
    if args.jsd_loss:
        assert num_aug_splits > 1  # JSD only valid with aug splits set
        train_loss_fn = JsdCrossEntropy(num_splits=num_aug_splits, smoothing=args.smoothing)
    elif mixup_active:
        # smoothing is handled with mixup target transform which outputs sparse, soft targets
        if args.bce_loss:
            train_loss_fn = BinaryCrossEntropy(target_threshold=args.bce_target_thresh)
        else:
            train_loss_fn = SoftTargetCrossEntropy()
    elif args.smoothing:
        if args.bce_loss:
            train_loss_fn = BinaryCrossEntropy(smoothing=args.smoothing, target_threshold=args.bce_target_thresh)
        else:
            train_loss_fn = LabelSmoothingCrossEntropy(smoothing=args.smoothing)
    else:
        train_loss_fn = nn.CrossEntropyLoss()
    train_loss_fn = train_loss_fn.to(device=device)
    validate_loss_fn = nn.CrossEntropyLoss().to(device=device)

    # setup checkpoint saver and eval metric tracking
    eval_metric = args.eval_metric
    best_metric = None
    best_epoch = None
    saver = None
    output_dir = None
    if utils.is_primary(args):
        if args.experiment:
            exp_name = args.experiment
        else:
            exp_name = '-'.join([
                datetime.now().strftime("%Y%m%d-%H%M%S"),
                safe_model_name(args.model),
                str(data_config['input_size'][-1])
            ])
        output_dir = utils.get_outdir(args.output if args.output else './output/train', exp_name)

        if has_synetune and args.report_synetune:
            #report = Reporter()
            output_dir = args.st_checkpoint_dir
            os.makedirs(output_dir, exist_ok=True)

        else:

            report = None

        decreasing = True if eval_metric == 'loss' else False
        saver = utils.CheckpointSaver(
            model=model,
            optimizer=optimizer,
            args=args,
            model_ema=model_ema,
            amp_scaler=loss_scaler,
            checkpoint_dir=output_dir,
            recovery_dir=output_dir,
            decreasing=decreasing,
            max_history=args.checkpoint_hist
        )
        with open(os.path.join(output_dir, 'args.yaml'), 'w') as f:
            f.write(args_text)

    # setup learning rate schedule and starting epoch
    updates_per_epoch = len(loader_train)
    lr_scheduler, num_epochs = create_scheduler_v2(
        optimizer,
        **scheduler_kwargs(args),
        updates_per_epoch=updates_per_epoch,
    )
    start_epoch = 0
    if args.start_epoch is not None:
        # a specified start_epoch will always override the resume epoch
        start_epoch = args.start_epoch
    elif resume_epoch is not None:
        start_epoch = resume_epoch
    if lr_scheduler is not None and start_epoch > 0:
        if args.sched_on_updates:
            lr_scheduler.step_update(start_epoch * updates_per_epoch)
        else:
            lr_scheduler.step(start_epoch)

    if args.initial_test:
        eval_metrics = validate(
            model,
            loader_eval,
            validate_loss_fn,
            args,
            amp_autocast=amp_autocast,
            return_features=return_features,
            return_source_output=return_source_output
        )
    else:
        eval_metrics = {"loss": -1,
                       "top1": -1,
                       "top5": -1}

    if utils.is_primary(args):
        _logger.info(
            f'Scheduled epochs: {num_epochs}. LR stepped per {"epoch" if lr_scheduler.t_in_epochs else "update"}.')

        initial_general_log = {"initial_eval_loss": eval_metrics['loss'],
                       "initial_eval_top1": eval_metrics['top1'],
                       "initial_eval_top5": eval_metrics['top5'],
                       "device_count": device_count}

        with open(os.path.join(output_dir, "initial_general_log.yml"), "w") as f:
            yaml.dump(initial_general_log, f, default_flow_style=False)

        #if device_count > 1: model = nn.DataParallel(model).to(device)
        invalid_loss_value = False



    try:
        for epoch in range(start_epoch, num_epochs):
            global_epoch = epoch
            if args.epochs_step !=-1 and epoch == args.epochs_step:
                break

            utils.random_seed(args.seed + epoch, args.rank)
            if hasattr(dataset_train, 'set_epoch'):
                dataset_train.set_epoch(epoch)
            elif args.distributed and hasattr(loader_train.sampler, 'set_epoch'):
                loader_train.sampler.set_epoch(epoch)

            if args.linear_probing and (epoch == (num_epochs-start_epoch)//2):
                model, head_name = prepare_model_for_finetuning(model, pct_to_freeze=args.pct_to_freeze,
                                                                return_features=return_features,
                                                                return_source_output=return_source_output,
                                                                change_head=False)
                #if device_count>1: model = nn.DataParallel(model).to(device)
            train_metrics = train_one_epoch(
                epoch=epoch,
                model=model,
                loader=loader_train,
                optimizer=optimizer,
                loss_fn=train_loss_fn,
                args=args,
                device=device,
                lr_scheduler=lr_scheduler,
                saver=saver,
                output_dir=output_dir,
                amp_autocast=amp_autocast,
                loss_scaler=loss_scaler,
                model_ema=model_ema,
                mixup_fn=mixup_fn,
                return_features=return_features,
                return_source_output=return_source_output,
                relationship=relationship,
                backbone_regularizations=backbone_regularizations,
                head_name=head_name
            )

            if args.distributed and args.dist_bn in ('broadcast', 'reduce'):
                if utils.is_primary(args):
                    _logger.info("Distributing BatchNorm running means and vars")
                utils.distribute_bn(model, args.world_size, args.dist_bn == 'reduce')

            eval_metrics = validate(
                model,
                loader_eval,
                validate_loss_fn,
                args,
                amp_autocast=amp_autocast,
                return_features=return_features,
                return_source_output=return_source_output
            )

            if has_synetune and args.report_synetune:
                report(epoch = epoch + 1, eval_accuracy = eval_metrics["top1"] )

            if model_ema is not None and not args.model_ema_force_cpu:
                if args.distributed and args.dist_bn in ('broadcast', 'reduce'):
                    utils.distribute_bn(model_ema, args.world_size, args.dist_bn == 'reduce')

                ema_eval_metrics = validate(
                    model_ema.module,
                    loader_eval,
                    validate_loss_fn,
                    args,
                    amp_autocast=amp_autocast,
                    log_suffix=' (EMA)',
                    return_features=return_features,
                    return_source_output=return_source_output
                )
                eval_metrics = ema_eval_metrics

            if output_dir is not None:
                lrs = [param_group['lr'] for param_group in optimizer.param_groups]
                utils.update_summary(
                    epoch,
                    train_metrics,
                    eval_metrics,
                    filename=os.path.join(output_dir, 'summary.csv'),
                    lr=sum(lrs) / len(lrs),
                    write_header=best_metric is None,
                    log_wandb=args.log_wandb and has_wandb,
                )

            if not args.test_mode and saver is not None:
                # save proper checkpoint with eval metric
                save_metric = eval_metrics[eval_metric]
                best_metric, best_epoch = saver.save_checkpoint(epoch, metric=save_metric)

            if lr_scheduler is not None:
                # step LR for next epoch
                lr_scheduler.step(epoch + 1, eval_metrics[eval_metric])

            if np.isnan(train_metrics["loss"]):
                invalid_loss_value = True
                break

    except KeyboardInterrupt:
        pass

    if best_metric is not None:
        _logger.info('*** Best metric: {0} (epoch {1})'.format(best_metric, best_epoch))

    if args.test_mode:

        with open(os.path.join(current_dir, "experiments", "results.txt"), 'a+') as f:
            #with open(f"experiments/results.txt", 'a+') as f:
            if not invalid_loss_value:
                f.write(f"ok {args.experiment}\n")
            else:
                f.write(f"Diverged {args.experiment}\n")

    if utils.is_primary(args):
        final_general_log = { "max_memory_allocated": torch.cuda.max_memory_allocated(),
                             "invalid_loss_value": invalid_loss_value}

        with open(os.path.join(output_dir, "final_general_log.yml"), "w") as f:
            yaml.dump(final_general_log, f, default_flow_style=False)

@extend_metrics
def train_one_epoch(
        epoch,
        model,
        loader,
        optimizer,
        loss_fn,
        args,
        device=torch.device('cuda'),
        lr_scheduler=None,
        saver=None,
        output_dir=None,
        amp_autocast=suppress,
        loss_scaler=None,
        model_ema=None,
        mixup_fn=None,
        return_features=False,
        return_source_output=False,
        relationship=None,
        backbone_regularizations=[],
        head_name=None):
    if args.mixup_off_epoch and epoch >= args.mixup_off_epoch:
        if args.prefetcher and loader.mixup_enabled:
            loader.mixup_enabled = False
        elif mixup_fn is not None:
            mixup_fn.mixup_enabled = False

    second_order = hasattr(optimizer, 'is_second_order') and optimizer.is_second_order
    batch_time_m = utils.AverageMeter()
    data_time_m = utils.AverageMeter()
    losses_m = utils.AverageMeter()
    head_grad_norm = 0.0
    backbone_grad_norm = 0.0
    num_logs = 0.0

    model.train()
    end = time.time()
    num_batches_per_epoch = len(loader)
    last_idx = num_batches_per_epoch - 1
    num_updates = epoch * num_batches_per_epoch


    for batch_idx, (input, target) in enumerate(loader):
        if args.test_mode:
            if batch_idx > 1:
                print("--------- test_mode set to True: breaking epoch for test reasons ---------")
                break
            
        last_batch = batch_idx == last_idx
        data_time_m.update(time.time() - end)
        if not args.prefetcher:
            input, target = input.to(device), target.to(device)
            if mixup_fn is not None:
                input, target = mixup_fn(input, target)
        if args.channels_last:
            input = input.contiguous(memory_format=torch.channels_last)

        with amp_autocast():
            output = model(input)
            if return_features:
                if return_source_output:
                    output, source_output, features = output
                else:
                    output, features = output

            loss = loss_fn(output, target)
            #print("0.", loss)
            for backbone_regularization in backbone_regularizations:
                if isinstance(backbone_regularization, SPRegularization):
                    loss += backbone_regularization()
                if isinstance(backbone_regularization, BatchSpectralShrinkage):
                    loss += backbone_regularization(features)
                if isinstance(backbone_regularization, BehavioralRegularization):
                    output = backbone_regularization.source_model(input)
                    if len(output) == 2:
                        output_source, layer_outputs_source = output
                    elif len(output) == 3:
                        output_source, _, layer_outputs_source = output
                    loss += backbone_regularization(layer_outputs_source, features)
                if isinstance(backbone_regularization, CoTuningLoss):
                    source_label = torch.from_numpy(relationship[target.cpu().numpy()]).to(device).float()
                    loss += backbone_regularization(source_output, source_label)

        if not args.distributed:
            losses_m.update(loss.item(), input.size(0))
        optimizer.zero_grad()
        if loss_scaler is not None:
            loss_scaler(
                loss, optimizer,
                clip_grad=args.clip_grad,
                clip_mode=args.clip_mode,
                parameters=model_parameters(model, exclude_head='agc' in args.clip_mode),
                create_graph=second_order
            )
        else:
            loss.backward(create_graph=second_order)
            if args.clip_grad is not None:
                utils.dispatch_clip_grad(
                    model_parameters(model, exclude_head='agc' in args.clip_mode),
                    value=args.clip_grad,
                    mode=args.clip_mode
                )
            optimizer.step()

        if model_ema is not None:
            model_ema.update(model)

        if device.type == 'cuda':
            torch.cuda.synchronize()
        num_updates += 1
        batch_time_m.update(time.time() - end)
        if last_batch or batch_idx % args.log_interval == 0:
            num_logs += 1
            lrl = [param_group['lr'] for param_group in optimizer.param_groups]
            lr = sum(lrl) / len(lrl)
            temp_backbone_grad_norm, temp_head_grad_norm = compute_gradient_norm(model, head_name)
            backbone_grad_norm = (num_logs-1)*backbone_grad_norm/num_logs + (1/num_logs)*temp_backbone_grad_norm
            head_grad_norm = (num_logs-1)*head_grad_norm/num_logs + (1/num_logs)*temp_head_grad_norm

            if args.distributed:
                reduced_loss = utils.reduce_tensor(loss.data, args.world_size)
                losses_m.update(reduced_loss.item(), input.size(0))

            if utils.is_primary(args):
                #fixing the batch_idx and last_idx

                _logger.info(
                    'Train: {} [{:>4d}/{} ({:>3.0f}%)]  '
                    'Loss: {loss.val:#.4g} ({loss.avg:#.3g})  '
                    'Time: {batch_time.val:.3f}s, {rate:>7.2f}/s  '
                    '({batch_time.avg:.3f}s, {rate_avg:>7.2f}/s)  '
                    'LR: {lr:.3e}  '
                    'BackboneGradNorm: {backbone_grad_norm:.3e}  '
                    'HeadGradNorm: {head_grad_norm:.3e}  '
                    'Data: {data_time.val:.3f} ({data_time.avg:.3f})'.format(
                        epoch,
                        batch_idx+1, len(loader),
                        100. * (batch_idx + 1)/ len(loader),
                        loss=losses_m,
                        batch_time=batch_time_m,
                        rate=input.size(0) * args.world_size / batch_time_m.val,
                        rate_avg=input.size(0) * args.world_size / batch_time_m.avg,
                        lr=lr,
                        data_time=data_time_m,
                        head_grad_norm=head_grad_norm,
                        backbone_grad_norm=backbone_grad_norm)
                )

                if args.save_images and output_dir:
                    torchvision.utils.save_image(
                        input,
                        os.path.join(output_dir, 'train-batch-%d.jpg' % batch_idx),
                        padding=0,
                        normalize=True
                    )

        if saver is not None and args.recovery_interval and (
                last_batch or (batch_idx + 1) % args.recovery_interval == 0):
            saver.save_recovery(epoch, batch_idx=batch_idx)

        if lr_scheduler is not None:
            lr_scheduler.step_update(num_updates=num_updates, metric=losses_m.avg)

        end = time.time()

        if np.isnan(losses_m.avg):
            break
        # end for

    if hasattr(optimizer, 'sync_lookahead'):
        optimizer.sync_lookahead()

    return OrderedDict([('loss', losses_m.avg), ('head_grad_norm', head_grad_norm), ('backbone_grad_norm', backbone_grad_norm)])

@extend_metrics
def validate(
        model,
        loader,
        loss_fn,
        args,
        device=torch.device('cuda'),
        amp_autocast=suppress,
        log_suffix='',
        return_features=False,
        return_source_output=False,
):
    batch_time_m = utils.AverageMeter()
    losses_m = utils.AverageMeter()
    top1_m = utils.AverageMeter()
    top5_m = utils.AverageMeter()

    model.eval()

    end = time.time()
    last_idx = len(loader) - 1
    with torch.no_grad():
        for batch_idx, (input, target) in enumerate(loader):
            if args.test_mode:
                if batch_idx > 1:
                    print("--------- test_mode set to True: breaking epoch for test reasons ---------")
                    break
            
            last_batch = batch_idx == last_idx
            if not args.prefetcher:
                input = input.to(device)
                target = target.to(device)
            if args.channels_last:
                input = input.contiguous(memory_format=torch.channels_last)

            with amp_autocast():
                output = model(input)
                if return_features:
                    if return_source_output:
                        output, _, _ = output
                    else:
                        output, _,  = output
            if isinstance(output, (tuple, list)):
                output = output[0]

            # augmentation reduction
            reduce_factor = args.tta
            if reduce_factor > 1:
                output = output.unfold(0, reduce_factor, reduce_factor).mean(dim=2)
                target = target[0:target.size(0):reduce_factor]

            loss = loss_fn(output, target)
            acc1, acc5 = utils.accuracy(output, target, topk=(1, 5))

            if args.distributed:
                reduced_loss = utils.reduce_tensor(loss.data, args.world_size)
                acc1 = utils.reduce_tensor(acc1, args.world_size)
                acc5 = utils.reduce_tensor(acc5, args.world_size)
            else:
                reduced_loss = loss.data

            if device.type == 'cuda':
                torch.cuda.synchronize()

            losses_m.update(reduced_loss.item(), input.size(0))
            top1_m.update(acc1.item(), output.size(0))
            top5_m.update(acc5.item(), output.size(0))

            batch_time_m.update(time.time() - end)
            end = time.time()
            if utils.is_primary(args) and (last_batch or batch_idx % args.log_interval == 0):
                log_name = 'Test' + log_suffix
                _logger.info(
                    '{0}: [{1:>4d}/{2}]  '
                    'Time: {batch_time.val:.3f} ({batch_time.avg:.3f})  '
                    'Loss: {loss.val:>7.4f} ({loss.avg:>6.4f})  '
                    'Acc@1: {top1.val:>7.4f} ({top1.avg:>7.4f})  '
                    'Acc@5: {top5.val:>7.4f} ({top5.avg:>7.4f})'.format(
                        log_name, batch_idx, last_idx,
                        batch_time=batch_time_m,
                        loss=losses_m,
                        top1=top1_m,
                        top5=top5_m)
                )
    metrics = OrderedDict([('loss', losses_m.avg), ('top1', top1_m.avg), ('top5', top5_m.avg)])

    return metrics


def make_one_batch_test(args, args_text, output_dir):
    args.test_mode = True
    args.linear_probing = False
    epochs = args.epochs
    torch.cuda.empty_cache()
    main(args, args_text)
    args.epochs = epochs
    args.test_mode = False
    args.linear_probing = True
    shutil.rmtree(output_dir)


if __name__ == '__main__':

    batch_size = None
    output_dir = None
    out_of_memory_error = True
    decrease_batch_size = False
    report = None
    args, args_text = _parse_args()

    while out_of_memory_error:
        try:
            torch.cuda.empty_cache()
            if decrease_batch_size:
                utils.get_outdir(args.output if args.output else './output/train', args.experiment)

            if batch_size:
                args.batch_size = batch_size
            args_text = yaml.safe_dump(args.__dict__, default_flow_style=False)
            batch_size = args.batch_size // 2
            #Test if there is enough memory
            if args.linear_probing:
                make_one_batch_test(args, args_text, output_dir)

            torch.cuda.empty_cache()
            main(args, args_text)
            out_of_memory_error = False

        except Exception as e:
            if isinstance(e, RuntimeError) and "out of memory" in str(e) and batch_size>1:
                if output_dir and decrease_batch_size:
                    shutil.rmtree(output_dir)
            else:
                with open(os.path.join(current_dir, "experiments", "results.txt"), 'a+') as f:
                    f.write(f"xx {args.experiment} ({e})\n")
                #with open(f"experiments/results.txt", 'a+') as f:
                #    f.write(f"xx {args.experiment} ({e})\n")
                print(e)
                out_of_memory_error = False
            if has_synetune and args.report_synetune and report is not None:
                report(epoch = global_epoch+1, eval_accuracy = 0 )
        if not decrease_batch_size:
            out_of_memory_error = False