train.sh

#!/bin/bash
# Run script
# Settings of training & test for different tasks.
method="$1"
task=$(python3 config.py)
case "${task}" in
    "DIS5K") epochs=600 && val_last=50 && step=5 ;;
    "COD") epochs=150 && val_last=50 && step=5 ;;
    "HRSOD") epochs=150 && val_last=50 && step=5 ;;
    "General") epochs=250 && val_last=30 && step=2 ;;
    "General-2K") epochs=250 && val_last=30 && step=2 ;;
    "Matting") epochs=100 && val_last=30 && step=2 ;;
esac
testsets=NO     # Non-existing folder to skip.
# testsets=TE-COD10K   # for COD

# Train
devices=$2
nproc_per_node=$(echo ${devices%%,} | grep -o "," | wc -l)

to_be_distributed=`echo ${nproc_per_node} | awk '{if($e > 0) print "True"; else print "False";}'`

echo Training started at $(date)
if [ ${to_be_distributed} == "True" ]
then
    # Adapt the nproc_per_node by the number of GPUs. Give 8989 as the default value of master_port.
    echo "Multi-GPU mode received..."
    CUDA_VISIBLE_DEVICES=${devices} \
    torchrun --nproc_per_node $((nproc_per_node+1)) --master_port=${3:-8999} \
    train.py --ckpt_dir ckpt/${method} --epochs ${epochs} \
        --testsets ${testsets} \
        --dist ${to_be_distributed} \
        --resume xx/xx-epoch_244.pth
else
    echo "Single-GPU mode received..."
    CUDA_VISIBLE_DEVICES=${devices} \
    python train.py --ckpt_dir ckpt/${method} --epochs ${epochs} \
        --testsets ${testsets} \
        --dist ${to_be_distributed} \
        --resume xx/xx-epoch_244.pth
fi

echo Training finished at $(date)