Prediction fails with OOM

[constantinpape]

python D:/constantin_Gits/flamingo-tools/scripts/prediction/run_prediction_distance_unet.py -i D:\DATA\M_LR_000155_L\2_converted_stitching\M_LR_000155_L.n5 -k setup2/timepoint0/s0 -m D:/constantin_Gits/Models/First_UnetModel_forSGNs/checkpoints/model_s0 -o D:\DATA\M_LR_000155_L\Predictions\SGNSingleTile02_from_n5

I get a CUDA out of memory error:

torch.OutOfMemoryError: CUDA out of memory. Tried to allocate 273.38 GiB. GPU 0 has a total capacity of 31.99 GiB of which 0 bytes is free. Of the allocated memory 39.03 GiB is allocated by PyTorch, and 3.22 GiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting PYTORCH_CUDA_ALLOC_CONF=expandable_segments:True to avoid fragmentation. See documentation for Memory Management (https://pytorch.org/docs/stable/notes/cuda.html#environment-variables)

[constantinpape]

Likely due to wrong tile shape