Function validate

eval_linear.py:222–265

(val_loader, model, criterion)

def validate(val_loader, model, criterion):
    batch_time = AverageMeter()
    losses = AverageMeter()
    top1 = AverageMeter()
    top5 = AverageMeter()

    # switch to evaluate mode
    model.eval()
    softmax = nn.Softmax(dim=1).cuda()
    end = time.time()
    with torch.no_grad():
        for i, (input, target) in enumerate(val_loader):
            if args.tencrops:
                bs, ncrops, c, h, w = input.size()
                input = input.view(-1, c, h, w).contiguous()
            input_cu, target_cu = input.cuda(), target.cuda()
            
            output = model(input_cu)
            if args.tencrops:
                output_central = output.view(bs, ncrops, -1)[: , int(ncrops / 2 - 1), :]
                output = softmax(output)
                output = torch.squeeze(output.view(bs, ncrops, -1).mean(1))
            else:
                output_central = output

            prec1, prec5 = accuracy(output.data, target_cu, topk=(1, 5))
            top1.update(prec1.item(), input.size(0))
            top5.update(prec5.item(), input.size(0))
            loss = criterion(output_central, target_cu)
            losses.update(loss.item(), input.size(0))

            # measure elapsed time
            batch_time.update(time.time() - end)
            end = time.time()

    logger.info('Validation: [{0}/{1}]\t'
                'Time {batch_time.val:.3f} ({batch_time.avg:.3f})\t'
                'Loss {loss.val:.4f} ({loss.avg:.4f})\t'
                'Prec@1 {top1.val:.3f} ({top1.avg:.3f})\t'
                'Prec@5 {top5.val:.3f} ({top5.avg:.3f})'
                .format(i, len(val_loader), batch_time=batch_time,
                 loss=losses, top1=top1, top5=top5))

    return top1.avg, top5.avg, losses.avg

if __name__ == '__main__':
    main()