nnUNet_train error, unable to write to file

[goodsave]

############################################### 2024-01-22 15:42:43.972844: Using dummy2d data augmentation loading dataset loading all case properties unpacking dataset done 2024-01-22 15:44:57.731584: lr: 0.01 using pin_memory on device 0 Traceback (most recent call last): File "/usr/lib/python3.6/multiprocessing/queues.py", line 234, in _feed obj = _ForkingPickler.dumps(obj) File "/usr/lib/python3.6/multiprocessing/reduction.py", line 51, in dumps cls(buf, protocol).dump(obj) File "/usr/local/lib/python3.6/dist-packages/torch/multiprocessing/reductions.py", line 328, in reduce_storage fd, size = storage._sharefd() RuntimeError: unable to write to file Traceback (most recent call last): File "/usr/lib/python3.6/multiprocessing/queues.py", line 234, in _feed obj = _ForkingPickler.dumps(obj) File "/usr/lib/python3.6/multiprocessing/reduction.py", line 51, in dumps cls(buf, protocol).dump(obj) File "/usr/local/lib/python3.6/dist-packages/torch/multiprocessing/reductions.py", line 328, in reduce_storage fd, size = storage._sharefd() RuntimeError: unable to write to file

[Kobalt93]

Hi goodsave, Could you please provide more details regarding your settings? It is hard to tell the problem without additional details. Could you please check whether

• you set your paths correctly (writing permissions?),
• your RAM has enough space, and
• The SingleThreadedAugmenter works? If you run your code in a docker container pleasecheck this thread too: https://github.com/MIC-DKFZ/nnUNet/issues/398