# default MODEL_NAME=convnext_base_in22k MODEL_PATH=../output/pretrained/DRCT-2M/sdv14/convnext_base_in22k_224_drct_amp_crop/14_acc0.9996.pth DEVICE_ID=0 EMBEDDING_SIZE=1024 MODEL_NAME=${1:-$MODEL_NAME} MODEL_PATH=${2:-$MODEL_PATH} DEVICE_ID=${3:-$DEVICE_ID} EMBEDDING_SIZE=${4:-$EMBEDDING_SIZE} ROOT_PATH=/disk1/chenby/dataset/AIGC_data/DRCT_data/MSCOCO/ FAKE_ROOT_PATH=/disk1/chenby/dataset/AIGC_data/DRCT_data/DRCT-2M/ DATASET_NAME=DRCT-2M SAVE_TXT=../output/results/DRCT-2M_metrics.txt INPUT_SIZE=224 BATCH_SIZE=24 FAKE_INDEXES=(1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16) for FAKE_INDEX in ${FAKE_INDEXES[@]} do echo FAKE_INDEX:${FAKE_INDEX} python train.py --root_path ${ROOT_PATH} --fake_root_path ${FAKE_ROOT_PATH} --model_name ${MODEL_NAME} \ --input_size ${INPUT_SIZE} --batch_size ${BATCH_SIZE} --device_id ${DEVICE_ID} --is_test \ --model_path ${MODEL_PATH} --is_crop --fake_indexes ${FAKE_INDEX} \ --save_txt ${SAVE_TXT} --embedding_size ${EMBEDDING_SIZE} done