run_random.sh

#!/bin/bash
# remember to add --save when you want to save the experiment log

GPU=$1

DATASET=$3
MODELS=$2
ARGS="$4 $5 $6"
# edge sampling ratio
SAMPLES=(1.0)
# number of training instances per class
TN=("20")
# hidden dimensions for graph revision module
HIDG=("100:10")

SEED_ARRAY=(633 684 806 350 208 857 369 153 329 363)
DATASEED_ARRAY=(103 219 977 678 527 549 368 945 573 920)

if [ ${MODELS} == 'GRCN' ]; then
    if [ ${DATASET} == "Cora" ]; then
        PARA=("30")
    elif [ ${DATASET} == "CiteSeer" ]; then
        PARA=("100")
    elif [ ${DATASET} == "PubMed" ]; then
        PARA=("5")
    elif [ ${DATASET} == "CoraFull" ]; then
        PARA=("20")
    elif [ ${DATASET} == "Computers" ]; then
        PARA=("30")
    elif [ ${DATASET} == "CS" ]; then
        PARA=("5")
    fi
elif [[ ${MODELS} == *"GRCN"* ]]; then
    if [ ${DATASET} == "Cora" ]; then
        PARA=("30")
    elif [ ${DATASET} == "CiteSeer" ]; then
        PARA=("100")
    elif [ ${DATASET} == "PubMed" ]; then
        PARA=("5")
    elif [ ${DATASET} == "CoraFull" ]; then
        PARA=("20")
    elif [ ${DATASET} == "Computers" ]; then
        PARA=("30")
    elif [ ${DATASET} == "CS" ]; then
        PARA=("5")
    fi
fi


for tn in "${TN[@]}"
do
    for sample in "${SAMPLES[@]}"
    do
        for para in "${PARA[@]}"
        do
            for hidg in "${HIDG[@]}"
            do
                for i in {0..9}
                do
                    SEED=${SEED_ARRAY[i]}
                    DATASEED=${DATASEED_ARRAY[i]}
                    option="--dataset ${DATASET} --sample ${sample} --complete ${MODELS}
                        --compl_param ${para} --seed ${SEED} --dataseed ${DATASEED} ${ARGS}
                        --hid_graph ${hidg} --train_num ${tn}"
                    cmd="CUDA_VISIBLE_DEVICES=${GPU} python main_ours.py ${option}"
                    echo $cmd
                    eval $cmd
                done
            done
        done
    done
done