Bug: RuntimeError: CUDA error: an illegal memory access was encountered

Description:

Если указать для forced alignment номер GPU отличный от 0, то появляется ошибка:

Traceback (most recent call last):
  File "/home/jovyan/popov_ilya/audio_framework/main.py", line 110, in <module>
    main()
  File "/home/jovyan/popov_ilya/audio_framework/main.py", line 83, in main
    processor()
  File "/home/jovyan/popov_ilya/audio_framework/src/preprocessing/aligner.py", line 324, in __call__
    sentence_spans = self.__compute_alignments(emission, texts)
                     ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/jovyan/popov_ilya/audio_framework/src/preprocessing/aligner.py", line 369, in __compute_alignments
    alignments, scores = forced_align(emission, targets, blank=self.config.blank_id)
                         ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/jovyan/popov_ilya/audio_framework/.venv/lib/python3.12/site-packages/torchaudio/functional/_alignment.py", line 72, in forced_align
    paths, scores = torch.ops.torchaudio.forced_align(log_probs, targets, input_lengths, target_lengths, blank)
                    ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/jovyan/popov_ilya/audio_framework/.venv/lib/python3.12/site-packages/torch/_ops.py", line 1061, in __call__
    return self_._op(*args, **(kwargs or {}))
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
RuntimeError: CUDA error: an illegal memory access was encountered
CUDA kernel errors might be asynchronously reported at some other API call, so the stacktrace below might be incorrect.
For debugging consider passing CUDA_LAUNCH_BLOCKING=1
Compile with `TORCH_USE_CUDA_DSA` to enable device-side assertions.

Solution

Для выбора определенной карты необходимо указать переменную окружения CUDA_VISIBLE_DEVICES

TODO

Добавить определение CUDA_VISIBLE_DEVICES в коде