ericyinyzy
commented
4 days ago Yes, because the algorithm need to search for adversarial text for each sample, which can not use the batch run command.
Open CharmingKillr opened 1 week ago
ericyinyzy
commented
4 days ago Yes, because the algorithm need to search for adversarial text for each sample, which can not use the batch run command.
Hello, when running attack_vqa under BLIP_attack, batch_size can only be set to 1, otherwise Traceback (most recent call last) will be reported: File "/opt/dpcvol/models/jw_projects/jw_projects/VLAttack/BLIP_attack/attack_vqa.py", line 140, in
main(args, config, config_pretrain)
File "/opt/dpcvol/models/jw_projects/jw_projects/VLAttack/BLIP_attack/attack_vqa.py", line 89, in main
pgd_attack.evaluate(test_loader, tokenizer)
The File "/ home/naie/local/lib/python3.9 / site - packages/torch/utils / _contextlib py", line 115, in decorate_context
return func(*args, **kwargs)
File "/opt/dpcvol/models/jw_projects/jw_projects/VLAttack/BLIP_attack/attack/VQA/adv_attack_blip.py", line 80, in evaluate
for i, batch in enumerate(metric_logger.log_every(data_loader, print_freq, header)):
File "/opt/dpcvol/models/jw_projects/jw_projects/VLAttack/BLIP_attack/utils.py", line 156, in log_every
for obj in iterable:
The File "/ home/naie/local/lib/python3.9 / site - packages/torch/utils/data/dataloader py", line 630, in next
data = self._next_data()
The File "/ home/naie/local/lib/python3.9 / site - packages/torch/utils/data/dataloader py", line 1345, in _next_data
return self._process_data(data)
The File "/ home/naie/local/lib/python3.9 / site - packages/torch/utils/data/dataloader py", line 1371, in _process_data
data.reraise()
The File "/ home/naie/local/lib/python3.9 / site - packages/torch / _utils. Py", line 694, in reraise
raise exception
RuntimeError: Caught RuntimeError in DataLoader worker process 0.
Original Traceback (most recent call last):
The File "/ home/naie/local/lib/python3.9 / site - packages/torch/utils/data / _utils/worker. Py", line 308, in _worker_loop
data = fetcher.fetch(index)
The File "/ home/naie/local/lib/python3.9 / site - packages/torch/utils/data / _utils/fetch. Py", line 54, fetch in
return self.collate_fn(data)
The File "/ home/naie/local/lib/python3.9 / site - packages/torch/utils/data / _utils/collate. Py", line 265, in default_collate
return collate(batch, collate_fn_map=default_collate_fn_map)
The File "/ home/naie/local/lib/python3.9 / site - packages/torch/utils/data / _utils/collate. Py", line 127, in collate
return elem_type({key: collate([d[key] for d in batch], collate_fn_map=collate_fn_map) for key in elem})
The File "/ home/naie/local/lib/python3.9 / site - packages/torch/utils/data / _utils/collate. Py", line 127, in < dictcomp >
return elem_type({key: collate([d[key] for d in batch], collate_fn_map=collate_fn_map) for key in elem})
The File "/ home/naie/local/lib/python3.9 / site - packages/torch/utils/data / _utils/collate. Py", line 138, in collate
raise RuntimeError('each element in list of batch should be of equal size')
RuntimeError: each element in list of batch should be of equal size