#Training details export CUDA_VISIBLE_DEVICES=0,1,2,3,4,5,6,7 export FLAGS_fraction_of_gpu_memory_to_use=0.98 #ResNeXt152_64x4d python train.py \ --model=ResNeXt152_64x4d \ --batch_size=256 \ --total_images=1281167 \ --image_shape=3,224,224 \ --class_dim=1000 \ --lr_strategy=piecewise_decay \ --lr=0.1 \ --num_epochs=120 \ --model_save_dir=output/ \ --l2_decay=18e-5
