How to run bart-base with SAMsum dataset

[JxuHenry]

My running script:
if [ $# != 2 ] ; then
echo "USAGE: $0 COMMENT"
echo " e.g.: $0 condigsum 2"
exit 1;
fi

GPU=${2}
if [ ${GPU} -eq -1 ] ; then
GPU=0,1,2,3
fi
echo "running on GPU ${GPU}"

WARMUP_UPDATES=200
MAX_TOKENS=800
UPDATE_FREQ=32

LR=4e-05
TOTAL_NUM_UPDATES=1000
dataset='SAMSumInd'
PRETRAIN_PATH='/root/jxu/ConDigSum-main/train_sh/bart_base/model.pt' # [YOUR_OWN_PATH] # to be updated....

CHECKPOINT_PATH="checkpoint_${1}base$dataset" && mkdir -p $CHECKPOINT_PATH
CUDA_VISIBLE_DEVICES=${GPU} python ../fairseq_cli/train.py $dataset
--restore-file $PRETRAIN_PATH
--max-tokens $MAX_TOKENS
--no-last-checkpoints
--task translation
--co_window_size 14
--ma2_minwin 5
--ma2_maxwin 24
--ma2_step 1
--co_sample_truncate 2
--ma_sample_truncate 2
--co_loss_weight 0.005
--ma_loss_weight 0.0001
--source-lang source --target-lang target
--truncate-source
--layernorm-embedding
--share-all-embeddings
--share-decoder-input-output-embed
--reset-optimizer --reset-dataloader --reset-meters
--arch bart_base
--criterion label_smoothed_cross_entropy
--label-smoothing 0.1
--dropout 0.1 --attention-dropout 0.1
--weight-decay 0.01 --optimizer adam --adam-betas "(0.9, 0.999)" --adam-eps 1e-08
--clip-norm 0.1
--lr-scheduler polynomial_decay --lr $LR --total-num-update $TOTAL_NUM_UPDATES --warmup-updates $WARMUP_UPDATES
--update-freq $UPDATE_FREQ
--skip-invalid-size-inputs-valid-test
--find-unused-parameters
--ddp-backend=no_c10d
--required-batch-size-multiple 1
--no-epoch-checkpoints
--save-dir $CHECKPOINT_PATH
--seed 14632 >> ${CHECKPOINT_PATH}/log

got the following error