Customized Dataset Validation Accuracy

Hello @nttstar ,

Currently, I'm training my customized dataset by recognition/training.py, the parameters as following:
batch_size=64, ckpt=3, ctx_num=1, dataset='emore', dataset_path=' ./facial_workspace/training/output', frequent=20, image_channel=3, image_shape=[112, 112, 3], kvstore='device', loss='arcface', lr=0.1, lr_steps='100000,160000,220000', models_root='./models', mom=0.9, network='r100', num_classes=90374, per_batch_size=64, pretrained='', pretrained_epoch=0, rescale_threshold=0, val_targets=['lfw', 'cfp_fp', 'agedb_30'], verbose=2000, wd=0.0005

My question is :
Can I stop this training as I found the "Accuracy-Highest" (value is 0.71433 that is not high enough) has not been changed long time.