deeplabv3_res101.sh

python -m torch.distributed.launch --nproc_per_node=8 train_baseline.py \
    --model deeplabv3 \
    --backbone resnet101 \
    --data [your dataset path]/cityscapes/ \
    --save-dir [your directory path to store checkpoint files] \
    --log-dir [your directory path to store log files] \
    --gpu-id 0,1,2,3,4,5,6,7 \
    --pretrained-base [your pretrained-backbone path]/resnet101-imagenet.pth