train_search.py

import argparse
import logging
import math
import os
import random
import time
from copy import deepcopy
from pathlib import Path
from threading import Thread

import numpy as np
import torch.distributed as dist
import torch.nn as nn
import torch.nn.functional as F
import torch.optim as optim
import torch.optim.lr_scheduler as lr_scheduler
import torch.utils.data
import yaml
from torch.cuda import amp
from torch.nn.parallel import DistributedDataParallel as DDP
from torch.utils.tensorboard import SummaryWriter
from tqdm import tqdm

import test  # import test.py to get mAP after each epoch
from models.experimental import attempt_load
from models.yolo_search import Model, parse_model
from utils.autoanchor import check_anchors
from utils.datasets import create_dataloader_search, create_dataloader
from utils.general import labels_to_class_weights, increment_path, labels_to_image_weights, init_seeds, \
    fitness, strip_optimizer_search, get_latest_run, check_dataset, check_file, check_git_status, check_img_size, \
    check_requirements, print_mutation, set_logging, one_cycle, colorstr
from utils.google_utils import attempt_download
from utils.loss import ComputeLoss
from utils.plots import plot_images, plot_labels, plot_results, plot_evolution
from utils.torch_utils import EMA, ModelEMA, select_device, intersect_dicts, torch_distributed_zero_first, is_parallel

from architect import Architect

logger = logging.getLogger(__name__)


def train(hyp, opt, device, tb_writer=None, wandb=None):
    logger.info(colorstr('hyperparameters: ') + ', '.join(f'{k}={v}' for k, v in hyp.items()))
    save_dir, epochs, batch_size, total_batch_size, weights, rank = \
        Path(opt.save_dir), opt.epochs, opt.batch_size, opt.total_batch_size, opt.weights, opt.global_rank

    # Directories
    wdir = save_dir / 'weights'
    wdir.mkdir(parents=True, exist_ok=True)  # make dir
    geno_dir = save_dir / 'genotypes'
    geno_dir.mkdir(parents=True, exist_ok=True)  # make dir
    alpha_dir = save_dir / 'alphas'
    alpha_dir.mkdir(parents=True, exist_ok=True)  # make dir
    last = wdir / 'last.pt'
    best = wdir / 'best.pt'
    results_file = save_dir / 'results.txt'

    # Save run settings
    with open(save_dir / 'hyp.yaml', 'w') as f:
        yaml.dump(hyp, f, sort_keys=False)
    with open(save_dir / 'opt.yaml', 'w') as f:
        yaml.dump(vars(opt), f, sort_keys=False)

    # Configure
    plots = not opt.evolve  # create plots
    cuda = device.type != 'cpu'
    init_seeds(2 + rank)
    with open(opt.data) as f:
        data_dict = yaml.load(f, Loader=yaml.SafeLoader)  # data dict
    with torch_distributed_zero_first(rank):
        check_dataset(data_dict)  # check
    train_path = data_dict['train']
    test_path = data_dict['val']
    nc = 1 if opt.single_cls else int(data_dict['nc'])  # number of classes
    names = ['item'] if opt.single_cls and len(data_dict['names']) != 1 else data_dict['names']  # class names
    assert len(names) == nc, '%g names found for nc=%g dataset in %s' % (len(names), nc, opt.data)  # check

    # Model
    pretrained = weights.endswith('.pt')
    if pretrained:
        with torch_distributed_zero_first(rank):
            attempt_download(weights)  # download if not found locally
        ckpt = torch.load(weights, map_location=device)  # load checkpoint
        model = Model(opt.cfg or ckpt['model'].yaml, ch=3, nc=nc, anchors=hyp.get('anchors')).to(device)  # create
        exclude = ['anchor'] if (opt.cfg or hyp.get('anchors')) and not opt.resume else []  # exclude keys
        state_dict = ckpt['model'].float().state_dict()  # to FP32
        state_dict = intersect_dicts(state_dict, model.state_dict(), exclude=exclude)  # intersect
        model.load_state_dict(state_dict, strict=False)  # load
        logger.info('Transferred %g/%g items from %s' % (len(state_dict), len(model.state_dict()), weights))  # report
    else:
        model = Model(opt.cfg, ch=3, nc=nc, anchors=hyp.get('anchors')).to(device)  # create
#    model.update_arch_parameters() # Since to(device) will not update model._arch_parameters
#    for alpha in model.arch_parameters():
#        alpha.requires_grad_(True)

    # Freeze
    freeze = []  # parameter names to freeze (full or partial)
    for k, v in model.named_parameters():
        v.requires_grad = True  # train all layers
        if any(x in k for x in freeze):
            print('freezing %s' % k)
            v.requires_grad = False

    # Optimizer
#    nbs = 64  # nominal batch size
#    nbs = total_batch_size  # nominal batch size
    nbs = max(total_batch_size, 64)  # nominal batch size
    accumulate = max(round(nbs / total_batch_size), 1)  # accumulate loss before optimizing
    hyp['weight_decay'] *= total_batch_size * accumulate / nbs  # scale weight_decay
    logger.info(f"Scaled weight_decay = {hyp['weight_decay']}")

    pg0, pg1, pg2 = [], [], []  # optimizer parameter groups
    for k, v in model.named_modules():
        if hasattr(v, 'bias') and isinstance(v.bias, nn.Parameter):
            pg2.append(v.bias)  # biases
        if isinstance(v, nn.BatchNorm2d) and hasattr(v, 'weight') and isinstance(v.weight, nn.Parameter):
            pg0.append(v.weight)  # no decay
        elif hasattr(v, 'weight') and isinstance(v.weight, nn.Parameter):
            pg1.append(v.weight)  # apply decay
        elif hasattr(v, 'depth_weight') and isinstance(v.depth_weight, nn.Parameter):
            pg1.append(v.depth_weight)  # apply decay
            if hasattr(v, 'point_weight') and isinstance(v.point_weight, nn.Parameter): pg1.append(v.point_weight)  # apply decay
        elif hasattr(v, 'depth_weight1') and isinstance(v.depth_weight1, nn.Parameter):
            pg1.append(v.depth_weight1)  # apply decay
        elif hasattr(v, 'depth_weight2') and isinstance(v.depth_weight2, nn.Parameter):
            pg1.append(v.depth_weight2)  # apply decay

    tt = 0
    for w in model.parameters(): tt+=1
    assert (tt == len(pg0)+len(pg1)+len(pg2))
#    # Test genotype code
#    geno, model_yaml = model.genotype()
#    parse_model(model_yaml, [3])
#    print(geno)
#    assert 0

    if opt.adam:
        optimizer = optim.Adam(pg0, lr=hyp['lr0'], betas=(hyp['momentum'], 0.999))  # adjust beta1 to momentum
    else:
        optimizer = optim.SGD(pg0, lr=hyp['lr0'], momentum=hyp['momentum'], nesterov=True)

    optimizer.add_param_group({'params': pg1, 'weight_decay': hyp['weight_decay']})  # add pg1 with weight_decay
    optimizer.add_param_group({'params': pg2})  # add pg2 (biases)
    logger.info('Optimizer groups: %g .bias, %g conv.weight, %g other' % (len(pg2), len(pg1), len(pg0)))
    del pg0, pg1, pg2

    # Scheduler https://arxiv.org/pdf/1812.01187.pdf
    # https://pytorch.org/docs/stable/_modules/torch/optim/lr_scheduler.html#OneCycleLR
    if opt.linear_lr:
        lf = lambda x: (1 - x / (epochs - 1)) * (1.0 - hyp['lrf']) + hyp['lrf']  # linear
    else:
        lf = one_cycle(1, hyp['lrf'], epochs)  # cosine 1->hyp['lrf']
    scheduler = lr_scheduler.LambdaLR(optimizer, lr_lambda=lf)
    # plot_lr_scheduler(optimizer, scheduler, epochs)

    # Logging
    if rank in [-1, 0] and wandb and wandb.run is None:
        opt.hyp = hyp  # add hyperparameters
        wandb_run = wandb.init(config=opt, resume="allow",
                               project='YOLOv5' if opt.project == 'runs/train' else Path(opt.project).stem,
                               name=save_dir.stem,
                               entity=opt.entity,
                               id=ckpt.get('wandb_id') if 'ckpt' in locals() else None)
    loggers = {'wandb': wandb}  # loggers dict

    # EMA
#    ema = ModelEMA(model) if rank in [-1, 0] else None
    ema = EMA(model) if rank in [-1, 0] else None

    # Resume
    start_epoch, best_fitness = 0, 0.0
    if pretrained:
        # Optimizer
        if ckpt['optimizer'] is not None:
            optimizer.load_state_dict(ckpt['optimizer'])
            best_fitness = ckpt['best_fitness']

        # EMA
        if ema and ckpt.get('ema'):
#            ema.ema.load_state_dict(ckpt['ema'].float().state_dict())
            ema.shadow = ckpt['ema']
            ema.updates = ckpt['updates']

        # Results
        if ckpt.get('training_results') is not None:
            results_file.write_text(ckpt['training_results'])  # write results.txt

        # Epochs
        start_epoch = ckpt['epoch'] + 1
        if opt.resume:
            assert start_epoch > 0, '%s training to %g epochs is finished, nothing to resume.' % (weights, epochs)
        if epochs < start_epoch:
            logger.info('%s has been trained for %g epochs. Fine-tuning for %g additional epochs.' %
                        (weights, ckpt['epoch'], epochs))
            epochs += ckpt['epoch']  # finetune additional epochs

        del ckpt, state_dict

    # Image sizes
    gs = max(int(model.stride.max()), 32)  # grid size (max stride)
    nl = model.model[-1].nl  # number of detection layers (used for scaling hyp['obj'])
    imgsz, imgsz_test = [check_img_size(x, gs) for x in opt.img_size]  # verify imgsz are gs-multiples

    # DP mode
    if cuda and rank == -1 and torch.cuda.device_count() > 1:
        model = torch.nn.DataParallel(model)

    # SyncBatchNorm
    if opt.sync_bn and cuda and rank != -1:
        model = torch.nn.SyncBatchNorm.convert_sync_batchnorm(model).to(device)
        logger.info('Using SyncBatchNorm()')

    # DDP mode
    if cuda and rank != -1:
        model = DDP(model, device_ids=[opt.local_rank], output_device=opt.local_rank)

    # Trainloader
    dataloader, dataloader_val, dataset, dataset_val = create_dataloader_search(train_path, imgsz, batch_size, gs, opt,
                                            hyp=hyp, augment=True, cache=opt.cache_images, rect=opt.rect, rank=rank,
                                            world_size=opt.world_size, workers=opt.workers,
                                            image_weights=opt.image_weights, quad=opt.quad, prefix=colorstr('train: '), train_portion=opt.train_portion)
    mlc = np.concatenate(dataset.dataset.labels, 0)[:, 0].max()  # max label class. dataset is an instance of torch.utils.data.Subset
    nb = len(dataloader)  # number of batches
    assert mlc < nc, 'Label class %g exceeds nc=%g in %s. Possible class labels are 0-%g' % (mlc, nc, opt.data, nc - 1)

    # Process 0
    if rank in [-1, 0]:
        testloader = create_dataloader(test_path, imgsz_test, batch_size * 2, gs, opt,  # testloader
                                       hyp=hyp, cache=opt.cache_images and not opt.notest, rect=True, rank=-1,
                                       world_size=opt.world_size, workers=opt.workers,
                                       pad=0.5, prefix=colorstr('val: '))[0]

        if not opt.resume:
            labels = np.concatenate(dataset.dataset.labels, 0)
            c = torch.tensor(labels[:, 0])  # classes
            # cf = torch.bincount(c.long(), minlength=nc) + 1.  # frequency
            # model._initialize_biases(cf.to(device))
            if plots:
                plot_labels(labels, save_dir, loggers)
                if tb_writer:
                    tb_writer.add_histogram('classes', c, 0)

            # Anchors
            if not opt.noautoanchor:
                check_anchors(dataset.dataset, model=model, thr=hyp['anchor_t'], imgsz=imgsz)
#            model.half().float()  # pre-reduce anchor precision 

    # Model parameters
    hyp['box'] *= 3. / nl  # scale to layers
    hyp['cls'] *= nc / 80. * 3. / nl  # scale to classes and layers
    hyp['obj'] *= (imgsz / 640) ** 2 * 3. / nl  # scale to image size and layers
    model.nc = nc  # attach number of classes to model
    model.hyp = hyp  # attach hyperparameters to model
    model.gr = 1.0  # iou loss ratio (obj_loss = 1.0 or iou)
    model.class_weights = labels_to_class_weights(dataset.dataset.labels, nc).to(device) * nc  # attach class weights
    model.names = names

    # Start training
    t0 = time.time()
    nw = max(round(hyp['warmup_epochs'] * nb), 1000)  # number of warmup iterations, max(3 epochs, 1k iterations)
    # nw = min(nw, (epochs - start_epoch) / 2 * nb)  # limit warmup to < 1/2 of training
    maps = np.zeros(nc)  # mAP per class
    results = (0, 0, 0, 0, 0, 0, 0)  # P, R, mAP@.5, mAP@.5-.95, val_loss(box, obj, cls)
    scheduler.last_epoch = start_epoch - 1  # do not move
    scaler = amp.GradScaler(enabled=cuda)
    compute_loss = ComputeLoss(model)  # init loss class
    logger.info(f'Image sizes {imgsz} train, {imgsz_test} test\n'
                f'Using {dataloader.num_workers} dataloader workers\n'
                f'Logging results to {save_dir}\n'
                f'Starting training for {epochs} epochs...')

    # Architect for NAS
    architect = Architect(model, compute_loss, accumulate, device, opt, DDP=torch.cuda.device_count()>1)
    ori_model = model.module if is_parallel(model) else model
    ori_model.display_alphas()

#    torch.autograd.set_detect_anomaly(True)

    for epoch in range(start_epoch, epochs):  # epoch ------------------------------------------------------------------
        model.train()

        # Update image weights (optional)
        if opt.image_weights:
            # Generate indices
            if rank in [-1, 0]:
                cw = model.class_weights.cpu().numpy() * (1 - maps) ** 2 / nc  # class weights
                iw = labels_to_image_weights(dataset.dataset.labels, nc=nc, class_weights=cw)  # image weights
                dataset.indices = random.choices(range(len(dataset.indices)), weights=iw[:len(dataset.indices)], k=len(dataset.indices))  # rand weighted idx
                dataset_val.indices = random.choices(range(len(dataset_val.indices)), weights=iw[-len(dataset_val.indices):], k=len(dataset_val.indices))  # rand weighted idx
            # Broadcast if DDP
            if rank != -1:
                indices = (torch.tensor(dataset.indices) if rank == 0 else torch.zeros(len(dataset.indices))).int()
                dist.broadcast(indices, 0)
                if rank != 0:
                    dataset.indices = indices.cpu().numpy()
                indices = (torch.tensor(dataset_val.indices) if rank == 0 else torch.zeros(len(dataset_val.indices))).int()
                dist.broadcast(indices, 0)
                if rank != 0:
                    dataset_val.indices = indices.cpu().numpy()

        # Update mosaic border
        # b = int(random.uniform(0.25 * imgsz, 0.75 * imgsz + gs) // gs * gs)
        # dataset.mosaic_border = [b - imgsz, -b]  # height, width borders

        mloss = torch.zeros(4, device=device)  # mean losses
        if rank != -1:
            dataloader.sampler.set_epoch(epoch)
            dataloader_val.sampler.set_epoch(epoch)
        pbar = enumerate(dataloader)
        logger.info(('\n' + '%10s' * 8) % ('Epoch', 'gpu_mem', 'box', 'obj', 'cls', 'total', 'targets', 'img_size'))
        if rank in [-1, 0]:
            pbar = tqdm(pbar, total=nb)  # progress bar
        optimizer.zero_grad()

        def valid_generator():
            while True:
              for x, t, path, shape in dataloader_val:
                yield x, t, path, shape
        valid_gen = valid_generator()
        for i, (imgs, targets, paths, _) in pbar:  # batch -------------------------------------------------------------
            ni = i + nb * epoch  # number integrated batches (since train start)
            imgs = imgs.to(device, non_blocking=True).float() / 255.0  # uint8 to float32, 0-255 to 0.0-1.0

            # Warmup
            if ni <= nw:
                xi = [0, nw]  # x interp
                # model.gr = np.interp(ni, xi, [0.0, 1.0])  # iou loss ratio (obj_loss = 1.0 or iou)
                accumulate = max(1, np.interp(ni, xi, [1, nbs / total_batch_size]).round())
                for j, x in enumerate(optimizer.param_groups):
                    # bias lr falls from 0.1 to lr0, all other lrs rise from 0.0 to lr0
                    x['lr'] = np.interp(ni, xi, [hyp['warmup_bias_lr'] if j == 2 else 0.0, x['initial_lr'] * lf(epoch)])
                    if 'momentum' in x:
                        x['momentum'] = np.interp(ni, xi, [hyp['warmup_momentum'], hyp['momentum']])

            # Multi-scale
            if opt.multi_scale:
                sz = random.randrange(imgsz * 0.5, imgsz * 1.5 + gs) // gs * gs  # size
                sf = sz / max(imgs.shape[2:])  # scale factor
                if sf != 1:
                    ns = [math.ceil(x * sf / gs) * gs for x in imgs.shape[2:]]  # new shape (stretched to gs-multiple)
                    imgs = F.interpolate(imgs, size=ns, mode='bilinear', align_corners=False)

            # architect
            if epoch >= opt.search_warmup:
#              input_valid = imgs
#              target_valid = targets
              input_valid, target_valid, _, _ = next(valid_gen)
              input_valid = input_valid.to(device, non_blocking=True).float() / 255.0  # uint8 to float32, 0-255 to 0.0-1.0
              # Multi-scale
              if opt.multi_scale:
                  sz = random.randrange(imgsz * 0.5, imgsz * 1.5 + gs) // gs * gs  # size
                  sf = sz / max(input_valid.shape[2:])  # scale factor
                  if sf != 1:
                      ns = [math.ceil(x * sf / gs) * gs for x in input_valid.shape[2:]]  # new shape (stretched to gs-multiple)
                      input_valid = F.interpolate(input_valid, size=ns, mode='bilinear', align_corners=False)
              architect.step(input_valid, target_valid)

            # Forward
            with amp.autocast(enabled=cuda):
                pred = model(imgs)  # forward
                loss, loss_items = compute_loss(pred, targets.to(device))  # loss scaled by batch_size
                if rank != -1:
                    loss *= opt.world_size  # gradient averaged between devices in DDP mode
                if opt.quad:
                    loss *= 4.

            # Backward
#            scaler.scale(loss).backward()
            grads =  torch.autograd.grad(scaler.scale(loss), model.parameters(), grad_outputs=torch.ones_like(loss), allow_unused=True)
#            for idx, (name, p) in enumerate(model.named_parameters()):
#              if grads[idx] is None: print(name)
#              else: print(grads[idx])
#            assert 0
            for v, g in zip(model.parameters(), grads):
              if v.grad is None:
                if not (g is None):
                  v.grad = torch.autograd.Variable(g.data)
              else:
                if not (g is None):
                  v.grad.data.add_(g.data)

            # Optimize
            if ni % accumulate == 0:
                scaler.step(optimizer)  # optimizer.step
                scaler.update()
                optimizer.zero_grad()
                if ema:
#                    ema.update(model)
                    ema.update()

            # Print
            if rank in [-1, 0]:
                mloss = (mloss * i + loss_items) / (i + 1)  # update mean losses
                mem = '%.3gG' % (torch.cuda.memory_reserved() / 1E9 if torch.cuda.is_available() else 0)  # (GB)
                s = ('%10s' * 2 + '%10.4g' * 6) % (
                    '%g/%g' % (epoch, epochs - 1), mem, *mloss, targets.shape[0], imgs.shape[-1])
                pbar.set_description(s)

                # Plot
                if plots and ni < 3:
                    f = save_dir / f'train_batch{ni}.jpg'  # filename
                    Thread(target=plot_images, args=(imgs, targets, paths, f), daemon=True).start()
                    # if tb_writer:
                    #     tb_writer.add_image(f, result, dataformats='HWC', global_step=epoch)
                    #     tb_writer.add_graph(model, imgs)  # add model to tensorboard
                elif plots and ni == 10 and wandb:
                    wandb.log({"Mosaics": [wandb.Image(str(x), caption=x.name) for x in save_dir.glob('train*.jpg')
                                           if x.exists()]}, commit=False)

#            # test whether the model in architect and ema is updated
#            state_dict_m = model.state_dict()
#            state_dict_a = architect.model.state_dict()
#            state_dict_e = ema.model.state_dict()
#            for key in state_dict_m.keys():
#              print((state_dict_m[key]-state_dict_a[key]).sum())
#              print((state_dict_m[key]-state_dict_e[key]).sum())
#            model.display_alphas()

            # end batch ------------------------------------------------------------------------------------------------
        # DDP process 0 or single-GPU
        if rank in [-1, 0]:
          # save alpha
          alpha_file = os.path.join(alpha_dir, '%d.yaml'%epoch)
          alphas_yaml = {}
          state_dict = ori_model.state_dict()
          for key in state_dict.keys():
            if 'alpha' in key:
              alphas_yaml[key] = state_dict[key].data.cpu().numpy().tolist()
          with open(alpha_file, encoding='utf-8', mode='w') as f:
            try:
                yaml.dump(data=alphas_yaml, stream=f, allow_unicode=True)
            except Exception as e:
                print(e)
          
          geno, model_yaml = ori_model.genotype()
          # save genotype
          geno_file = os.path.join(geno_dir, '%d.yaml'%epoch)
          with open(geno_file, encoding='utf-8', mode='w') as f:
            try:
                yaml.dump(data=model_yaml, stream=f, allow_unicode=True)
            except Exception as e:
                print(e)
          print("==================")
          print("normalized alphas:")
          ori_model.display_alphas()
          print('genotype:')
          for g in geno:
            print(g)
          print("==================")
        
        # end epoch ----------------------------------------------------------------------------------------------------

        # Scheduler
        lr = [x['lr'] for x in optimizer.param_groups]  # for tensorboard
        scheduler.step()

        # DDP process 0 or single-GPU
        if rank in [-1, 0]:
            # mAP
#            ema.update_attr(model, include=['yaml', 'nc', 'hyp', 'gr', 'names', 'stride', 'class_weights'])
            final_epoch = epoch + 1 == epochs
            if not opt.notest or final_epoch:  # Calculate mAP
                ema.apply_shadow()
                results, maps, times = test.test(opt.data,
                                                 batch_size=batch_size * 2,
                                                 imgsz=imgsz_test,
#                                                 model=ema.ema,
                                                 model=model.module if is_parallel(model) else model,
                                                 single_cls=opt.single_cls,
                                                 dataloader=testloader,
                                                 save_dir=save_dir,
                                                 verbose=nc < 50 and final_epoch,
                                                 plots=plots and final_epoch,
                                                 log_imgs=opt.log_imgs if wandb else 0,
                                                 compute_loss=compute_loss)
                ema.restore()

            # Write
            with open(results_file, 'a') as f:
                f.write(s + '%10.4g' * 7 % results + '\n')  # append metrics, val_loss
            if len(opt.name) and opt.bucket:
                os.system('gsutil cp %s gs://%s/results/results%s.txt' % (results_file, opt.bucket, opt.name))

            # Log
            tags = ['train/box_loss', 'train/obj_loss', 'train/cls_loss',  # train loss
                    'metrics/precision', 'metrics/recall', 'metrics/mAP_0.5', 'metrics/mAP_0.5:0.95',
                    'val/box_loss', 'val/obj_loss', 'val/cls_loss',  # val loss
                    'x/lr0', 'x/lr1', 'x/lr2']  # params
            for x, tag in zip(list(mloss[:-1]) + list(results) + lr, tags):
                if tb_writer:
                    tb_writer.add_scalar(tag, x, epoch)  # tensorboard
                if wandb:
                    wandb.log({tag: x}, step=epoch, commit=tag == tags[-1])  # W&B

            # Update best mAP
            fi = fitness(np.array(results).reshape(1, -1))  # weighted combination of [P, R, mAP@.5, mAP@.5-.95]
            if fi > best_fitness:
                best_fitness = fi

            # Save model
            if (not opt.nosave) or (final_epoch and not opt.evolve):  # if save
                ckpt = {'epoch': epoch,
                        'best_fitness': best_fitness,
                        'training_results': results_file.read_text(),
#                        'model': (model.module if is_parallel(model) else model).half(),
                        'model': model.module if is_parallel(model) else model,
#                        'ema': deepcopy(ema.ema).half(),
                        'ema': ema.shadow,
                        'updates': ema.updates,
                        'optimizer': optimizer.state_dict(),
                        'wandb_id': wandb_run.id if wandb else None}
                # Save last, best and delete
                torch.save(ckpt, last)
                if best_fitness == fi:
                    torch.save(ckpt, best)
                del ckpt
                model.float()

        # end epoch ----------------------------------------------------------------------------------------------------
    # end training

    if rank in [-1, 0]:
        # Strip optimizers
        final = best if best.exists() else last  # final model
        for f in last, best:
            if f.exists():
                tmp = f.with_name('stripped_%s'%f.name)
                strip_optimizer_search(f, s=tmp)
        if opt.bucket:
            os.system(f'gsutil cp {final} gs://{opt.bucket}/weights')  # upload

        # Plots
        if plots:
            plot_results(save_dir=save_dir)  # save as results.png
            if wandb:
                files = ['results.png', 'confusion_matrix.png', *[f'{x}_curve.png' for x in ('F1', 'PR', 'P', 'R')]]
                wandb.log({"Results": [wandb.Image(str(save_dir / f), caption=f) for f in files
                                       if (save_dir / f).exists()]})
                if opt.log_artifacts:
                    wandb.log_artifact(artifact_or_path=str(final), type='model', name=save_dir.stem)

        # Test best.pt
        logger.info('%g epochs completed in %.3f hours.\n' % (epoch - start_epoch + 1, (time.time() - t0) / 3600))
#        best = best.with_name('stripped_%s'%best.name)
        if opt.data.endswith('coco.yaml') and nc == 80:  # if COCO
            for m in (last, best) if best.exists() else (last):  # speed, mAP tests
                stripped_m = m.with_name('stripped_%s'%m.name)
                print(stripped_m)
                results, _, _ = test.test(opt.data,
                                          batch_size=batch_size * 2,
                                          imgsz=imgsz_test,
                                          conf_thres=0.001,
                                          iou_thres=0.7,
#                                          model=attempt_load(stripped_m, device).half(),
                                          model=attempt_load(stripped_m, device),
                                          single_cls=opt.single_cls,
                                          dataloader=testloader,
                                          save_dir=save_dir,
                                          save_json=True,
                                          plots=False)

    else:
        dist.destroy_process_group()

    wandb.run.finish() if wandb and wandb.run else None
    torch.cuda.empty_cache()
    return results


if __name__ == '__main__':
    parser = argparse.ArgumentParser()
    parser.add_argument('--weights', type=str, default='yolov5s.pt', help='initial weights path')
    parser.add_argument('--cfg', type=str, default='', help='model.yaml path')
    parser.add_argument('--data', type=str, default='data/coco128.yaml', help='data.yaml path')
    parser.add_argument('--hyp', type=str, default='data/hyp.scratch.yaml', help='hyperparameters path')
    parser.add_argument('--epochs', type=int, default=300)
    parser.add_argument('--batch-size', type=int, default=16, help='total batch size for all GPUs')
    parser.add_argument('--img-size', nargs='+', type=int, default=[640, 640], help='[train, test] image sizes')
    parser.add_argument('--rect', action='store_true', help='rectangular training')
    parser.add_argument('--resume', nargs='?', const=True, default=False, help='resume most recent training')
    parser.add_argument('--nosave', action='store_true', help='only save final checkpoint')
    parser.add_argument('--notest', action='store_true', help='only test final epoch')
    parser.add_argument('--noautoanchor', action='store_true', help='disable autoanchor check')
    parser.add_argument('--evolve', action='store_true', help='evolve hyperparameters')
    parser.add_argument('--bucket', type=str, default='', help='gsutil bucket')
    parser.add_argument('--cache-images', action='store_true', help='cache images for faster training')
    parser.add_argument('--image-weights', action='store_true', help='use weighted image selection for training')
    parser.add_argument('--device', default='', help='cuda device, i.e. 0 or 0,1,2,3 or cpu')
    parser.add_argument('--multi-scale', action='store_true', help='vary img-size +/- 50%%')
    parser.add_argument('--single-cls', action='store_true', help='train multi-class data as single-class')
    parser.add_argument('--adam', action='store_true', help='use torch.optim.Adam() optimizer')
    parser.add_argument('--sync-bn', action='store_true', help='use SyncBatchNorm, only available in DDP mode')
    parser.add_argument('--local_rank', type=int, default=-1, help='DDP parameter, do not modify')
    parser.add_argument('--log-imgs', type=int, default=16, help='number of images for W&B logging, max 100')
    parser.add_argument('--log-artifacts', action='store_true', help='log artifacts, i.e. final trained model')
    parser.add_argument('--workers', type=int, default=8, help='maximum number of dataloader workers')
    parser.add_argument('--project', default='runs/train', help='save to project/name')
    parser.add_argument('--entity', default=None, help='W&B entity')
    parser.add_argument('--name', default='exp', help='save to project/name')
    parser.add_argument('--exist-ok', action='store_true', help='existing project/name ok, do not increment')
    parser.add_argument('--quad', action='store_true', help='quad dataloader')
    parser.add_argument('--linear-lr', action='store_true', help='linear LR')

    # For NAS
    parser.add_argument('--arch_learning_rate', type=float, default=3e-4, help='learning rate for arch encoding')
    parser.add_argument('--arch_weight_decay', type=float, default=1e-3, help='weight decay for arch encoding')
    parser.add_argument('--search_warmup', type=int, default=0, help='Epoch to Warmup the operation weights')
    parser.add_argument('--train_portion', type=float, default=0.5, help='portion to split the train set and search set')

    opt = parser.parse_args()
    opt.project = '{}-{}'.format(opt.project, time.strftime("%Y%m%d-%H%M%S"))
    print("Experiments dir: %s"%opt.project)
    print("cfg file: %s"%opt.cfg)

    # Set DDP variables
    opt.world_size = int(os.environ['WORLD_SIZE']) if 'WORLD_SIZE' in os.environ else 1
    opt.global_rank = int(os.environ['RANK']) if 'RANK' in os.environ else -1
    set_logging(opt.global_rank)
    if opt.global_rank in [-1, 0]:
        check_git_status()
        check_requirements()

    # Resume
    if opt.resume:  # resume an interrupted run
        ckpt = opt.resume if isinstance(opt.resume, str) else get_latest_run()  # specified or most recent path
        assert os.path.isfile(ckpt), 'ERROR: --resume checkpoint does not exist'
        apriori = opt.global_rank, opt.local_rank
        with open(Path(ckpt).parent.parent / 'opt.yaml') as f:
            opt = argparse.Namespace(**yaml.load(f, Loader=yaml.SafeLoader))  # replace
        opt.cfg, opt.weights, opt.resume, opt.batch_size, opt.global_rank, opt.local_rank = '', ckpt, True, opt.total_batch_size, *apriori  # reinstate
        logger.info('Resuming training from %s' % ckpt)
    else:
        # opt.hyp = opt.hyp or ('hyp.finetune.yaml' if opt.weights else 'hyp.scratch.yaml')
        opt.data, opt.cfg, opt.hyp = check_file(opt.data), check_file(opt.cfg), check_file(opt.hyp)  # check files
        assert len(opt.cfg) or len(opt.weights), 'either --cfg or --weights must be specified'
        opt.img_size.extend([opt.img_size[-1]] * (2 - len(opt.img_size)))  # extend to 2 sizes (train, test)
        opt.name = 'evolve' if opt.evolve else opt.name
        opt.save_dir = increment_path(Path(opt.project) / opt.name, exist_ok=opt.exist_ok | opt.evolve)  # increment run

    # DDP mode
    opt.total_batch_size = opt.batch_size
    device = select_device(opt.device, batch_size=opt.batch_size)
    if opt.local_rank != -1:
        assert torch.cuda.device_count() > opt.local_rank
        torch.cuda.set_device(opt.local_rank)
        device = torch.device('cuda', opt.local_rank)
        dist.init_process_group(backend='nccl', init_method='env://')  # distributed backend
        assert opt.batch_size % opt.world_size == 0, '--batch-size must be multiple of CUDA device count'
        opt.batch_size = opt.total_batch_size // opt.world_size

    # Hyperparameters
    with open(opt.hyp) as f:
        hyp = yaml.load(f, Loader=yaml.SafeLoader)  # load hyps

    # Train
    logger.info(opt)
    try:
        import wandb
    except ImportError:
        wandb = None
        prefix = colorstr('wandb: ')
        logger.info(f"{prefix}Install Weights & Biases for YOLOv5 logging with 'pip install wandb' (recommended)")
    if not opt.evolve:
        tb_writer = None  # init loggers
        if opt.global_rank in [-1, 0]:
            logger.info(f'Start Tensorboard with "tensorboard --logdir {opt.project}", view at http://localhost:6006/')
            tb_writer = SummaryWriter(opt.save_dir)  # Tensorboard
        train(hyp, opt, device, tb_writer, wandb)

    # Evolve hyperparameters (optional)
    else:
        # Hyperparameter evolution metadata (mutation scale 0-1, lower_limit, upper_limit)
        meta = {'lr0': (1, 1e-5, 1e-1),  # initial learning rate (SGD=1E-2, Adam=1E-3)
                'lrf': (1, 0.01, 1.0),  # final OneCycleLR learning rate (lr0 * lrf)
                'momentum': (0.3, 0.6, 0.98),  # SGD momentum/Adam beta1
                'weight_decay': (1, 0.0, 0.001),  # optimizer weight decay
                'warmup_epochs': (1, 0.0, 5.0),  # warmup epochs (fractions ok)
                'warmup_momentum': (1, 0.0, 0.95),  # warmup initial momentum
                'warmup_bias_lr': (1, 0.0, 0.2),  # warmup initial bias lr
                'box': (1, 0.02, 0.2),  # box loss gain
                'cls': (1, 0.2, 4.0),  # cls loss gain
                'cls_pw': (1, 0.5, 2.0),  # cls BCELoss positive_weight
                'obj': (1, 0.2, 4.0),  # obj loss gain (scale with pixels)
                'obj_pw': (1, 0.5, 2.0),  # obj BCELoss positive_weight
                'iou_t': (0, 0.1, 0.7),  # IoU training threshold
                'anchor_t': (1, 2.0, 8.0),  # anchor-multiple threshold
                'anchors': (2, 2.0, 10.0),  # anchors per output grid (0 to ignore)
                'fl_gamma': (0, 0.0, 2.0),  # focal loss gamma (efficientDet default gamma=1.5)
                'hsv_h': (1, 0.0, 0.1),  # image HSV-Hue augmentation (fraction)
                'hsv_s': (1, 0.0, 0.9),  # image HSV-Saturation augmentation (fraction)
                'hsv_v': (1, 0.0, 0.9),  # image HSV-Value augmentation (fraction)
                'degrees': (1, 0.0, 45.0),  # image rotation (+/- deg)
                'translate': (1, 0.0, 0.9),  # image translation (+/- fraction)
                'scale': (1, 0.0, 0.9),  # image scale (+/- gain)
                'shear': (1, 0.0, 10.0),  # image shear (+/- deg)
                'perspective': (0, 0.0, 0.001),  # image perspective (+/- fraction), range 0-0.001
                'flipud': (1, 0.0, 1.0),  # image flip up-down (probability)
                'fliplr': (0, 0.0, 1.0),  # image flip left-right (probability)
                'mosaic': (1, 0.0, 1.0),  # image mixup (probability)
                'mixup': (1, 0.0, 1.0)}  # image mixup (probability)

        assert opt.local_rank == -1, 'DDP mode not implemented for --evolve'
        opt.notest, opt.nosave = True, True  # only test/save final epoch
        # ei = [isinstance(x, (int, float)) for x in hyp.values()]  # evolvable indices
        yaml_file = Path(opt.save_dir) / 'hyp_evolved.yaml'  # save best result here
        if opt.bucket:
            os.system('gsutil cp gs://%s/evolve.txt .' % opt.bucket)  # download evolve.txt if exists

        for _ in range(300):  # generations to evolve
            if Path('evolve.txt').exists():  # if evolve.txt exists: select best hyps and mutate
                # Select parent(s)
                parent = 'single'  # parent selection method: 'single' or 'weighted'
                x = np.loadtxt('evolve.txt', ndmin=2)
                n = min(5, len(x))  # number of previous results to consider
                x = x[np.argsort(-fitness(x))][:n]  # top n mutations
                w = fitness(x) - fitness(x).min()  # weights
                if parent == 'single' or len(x) == 1:
                    # x = x[random.randint(0, n - 1)]  # random selection
                    x = x[random.choices(range(n), weights=w)[0]]  # weighted selection
                elif parent == 'weighted':
                    x = (x * w.reshape(n, 1)).sum(0) / w.sum()  # weighted combination

                # Mutate
                mp, s = 0.8, 0.2  # mutation probability, sigma
                npr = np.random
                npr.seed(int(time.time()))
                g = np.array([x[0] for x in meta.values()])  # gains 0-1
                ng = len(meta)
                v = np.ones(ng)
                while all(v == 1):  # mutate until a change occurs (prevent duplicates)
                    v = (g * (npr.random(ng) < mp) * npr.randn(ng) * npr.random() * s + 1).clip(0.3, 3.0)
                for i, k in enumerate(hyp.keys()):  # plt.hist(v.ravel(), 300)
                    hyp[k] = float(x[i + 7] * v[i])  # mutate

            # Constrain to limits
            for k, v in meta.items():
                hyp[k] = max(hyp[k], v[1])  # lower limit
                hyp[k] = min(hyp[k], v[2])  # upper limit
                hyp[k] = round(hyp[k], 5)  # significant digits

            # Train mutation
            results = train(hyp.copy(), opt, device, wandb=wandb)

            # Write mutation results
            print_mutation(hyp.copy(), results, yaml_file, opt.bucket)

        # Plot results
        plot_evolution(yaml_file)
        print(f'Hyperparameter evolution complete. Best results saved as: {yaml_file}\n'
              f'Command to train a new model with these hyperparameters: $ python train.py --hyp {yaml_file}')