🐛 Bug

I'am trying to run the trainning code follow the waitk guide file , and fixed some bug just as @ereday this issue mentioned , but still got error when i ran the train code :

RuntimeError: Output 0 of SplitBackward0 is a view and is being modified inplace. This view is the output of a function that returns multiple views. Such functions do not allow the output views to be modified inplace. You should replace the inplace operation by an out-of-place one.

To Reproduce

Steps to reproduce the behavior (always include the command you ran):

k=7
MODEL=tf_wait${k}_wmt14Ende
CUDA_VISIBLE_DEVICES=0 python train.py $DATA_BIN -s en -t de --left-pad-source False \
    --user-dir examples/waitk --arch waitk_transformer_small \
    --save-dir $Workdir/checkpoints/$MODEL --tensorboard-logdir $Workdir/logs/$MODEL \
    --seed 1 --no-epoch-checkpoints --no-progress-bar --log-interval 10  \
    --optimizer adam --adam-betas '(0.9, 0.98)' --weight-decay 0.0001 \
    --max-tokens 4000 --update-freq 2 --max-update 50000 \
    --lr-scheduler inverse_sqrt --warmup-updates 4000 --warmup-init-lr '1e-07' --lr 0.002 \
    --min-lr '1e-9' --criterion label_smoothed_cross_entropy --label-smoothing 0.1 \
    --share-decoder-input-output-embed --waitk  $k

2. See error

Expected behavior

Environment

• fairseq Version (e.g., 1.0 or master):
• PyTorch Version (e.g., 1.0)
• OS (e.g., Linux):
• How you installed fairseq (pip, source):
• Build command you used (if compiling from source):
• Python version:
• CUDA/cuDNN version:
• GPU models and configuration:
• Any other relevant information:

Additional context

this repo seems out-of-date and the issue raised half years ago is still no replied.