I have an .npz file where I have stored a dictionary. The dictionary has some keys and the values are numpy arrays. I want to read the dictionary in my getitem() method of the dataloader. When I set the dataloader num_workers to 1, everything runs fine. But when I increase the num workers, it throws the following error when reading the data from that npz file:
Traceback (most recent call last):
File "scripts/train.py", line 235, in <module>
train(args)
File "scripts/train.py", line 186, in train
solver(args.epoch, args.verbose)
File "/local-scratch/codebase/cap/lib/solver.py", line 174, in __call__
self._feed(self.dataloader["train"], "train", epoch_id)
File "/local-scratch/codebase/cap/lib/solver.py", line 366, in _feed
for data_dict in dataloader:
File "/local-scratch/anaconda3/envs/scanenv/lib/python3.6/site-packages/torch/utils/data/dataloader.py", line 819, in __next__
return self._process_data(data)
File "/local-scratch/anaconda3/envs/scanenv/lib/python3.6/site-packages/torch/utils/data/dataloader.py", line 846, in _process_data
data.reraise()
File "/local-scratch/anaconda3/envs/scanenv/lib/python3.6/site-packages/torch/_utils.py", line 369, in reraise
raise self.exc_type(msg)
zipfile.BadZipFile: Caught BadZipFile in DataLoader worker process 0.
Original Traceback (most recent call last):
File "/local-scratch/anaconda3/envs/scanenv/lib/python3.6/site-packages/torch/utils/data/_utils/worker.py", line 178, in _worker_loop
data = fetcher.fetch(index)
File "/local-scratch/anaconda3/envs/scanenv/lib/python3.6/site-packages/torch/utils/data/_utils/fetch.py", line 44, in fetch
data = [self.dataset[idx] for idx in possibly_batched_index]
File "/local-scratch/anaconda3/envs/scanenv/lib/python3.6/site-packages/torch/utils/data/_utils/fetch.py", line 44, in <listcomp>
data = [self.dataset[idx] for idx in possibly_batched_index]
File "/local-scratch/codebase/cap/lib/dataset.py", line 947, in __getitem__
other_bbox_feat = self.box_features['{}-{}_{}.{}'.format(scene_id, target_object_id, ann_id, object_id)]
File "/local-scratch/anaconda3/envs/scanenv/lib/python3.6/site-packages/numpy/lib/npyio.py", line 255, in __getitem__
pickle_kwargs=self.pickle_kwargs)
File "/local-scratch/anaconda3/envs/scanenv/lib/python3.6/site-packages/numpy/lib/format.py", line 763, in read_array
data = _read_bytes(fp, read_size, "array data")
File "/local-scratch/anaconda3/envs/scanenv/lib/python3.6/site-packages/numpy/lib/format.py", line 892, in _read_bytes
r = fp.read(size - len(data))
File "/local-scratch/anaconda3/envs/scanenv/lib/python3.6/zipfile.py", line 872, in read
data = self._read1(n)
File "/local-scratch/anaconda3/envs/scanenv/lib/python3.6/zipfile.py", line 962, in _read1
self._update_crc(data)
File "/local-scratch/anaconda3/envs/scanenv/lib/python3.6/zipfile.py", line 890, in _update_crc
raise BadZipFile("Bad CRC-32 for file %r" % self.name)
zipfile.BadZipFile: Bad CRC-32 for file 'scene0519_00-13_1.0.npy'
As far as I know, pytorch dataloader uses multiprocessing to for data loading. Perhaps the issue is with multiprocessing and .npz files. I really appreciate any help.