Issue arise training on HPC

[markllai]

HPC using SLURM and a100 gpu
and the docker is convert into apptainer in order to adapt into HPC

loading annotations into memory...
0:00:00.070761
creating index...
index created!
Loading and preparing results...
DONE (t=0.01s)
creating index...
index created!
tokenization...
Traceback (most recent call last):
File "src/tasks/run_caption_VidSwinBert.py", line 679, in
main(args)
File "src/tasks/run_caption_VidSwinBert.py", line 666, in main
train(args, train_dataloader, val_dataloader, vl_transformer, tokenizer, training_saver, optimizer, scheduler)
File "src/tasks/run_caption_VidSwinBert.py", line 277, in train
evaluate_file = evaluate(args, val_dataloader, model, tokenizer, checkpoint_dir)
File "src/tasks/run_caption_VidSwinBert.py", line 343, in evaluate
result = evaluate_on_coco_caption(predict_file, caption_file, outfile=evaluate_file)
File "/videocap/src/evalcap/utils_caption_evaluate.py", line 99, in evaluate_on_coco_caption
cocoEval.evaluate()
File "/videocap/src/evalcap/coco_caption/pycocoevalcap/eval.py", line 41, in evaluate
self.tokenize()
File "/videocap/src/evalcap/coco_caption/pycocoevalcap/eval.py", line 37, in tokenize
self.gts = tokenizer.tokenize(gts)
File "/videocap/src/evalcap/coco_caption/pycocoevalcap/tokenizer/ptbtokenizer.py", line 43, in tokenize
tmp_file = tempfile.NamedTemporaryFile(delete=False, dir=path_to_jar_dirname)
File "/opt/conda/lib/python3.8/tempfile.py", line 541, in NamedTemporaryFile
(fd, name) = _mkstemp_inner(dir, prefix, suffix, flags, output_type)
File "/opt/conda/lib/python3.8/tempfile.py", line 250, in _mkstemp_inner
fd = _os.open(file, flags, 0o600)
PermissionError: [Errno 13] Permission denied: '/videocap/src/evalcap/coco_caption/pycocoevalcap/tokenizer/tmp8215_vrp'