Support for multi-gpu setup?

[Prasoon-Rai]

Hey is it possible to run the model on a 2x T4 setup provided by Kaggle? If there are any workarounds possible kindly list 'em.

[loretoparisi]

That would be game changer! on the L4/24GB VRAM

torch.OutOfMemoryError: CUDA out of memory. Tried to allocate 160.00 MiB. GPU 0 has a total capacity of 21.95 GiB of which 154.12 MiB is free. Including non-PyTorch memory, this process has 0 bytes memory in use. Of the allocated memory 21.24 GiB is allocated by PyTorch, and 355.20 MiB 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)