Joe Wallwork

The multi-GPU example currently puts the input tensors and model on CUDA devices, but has the output tensors on the CPU. We should test keeping the output on the GPU.

Drop support for integer tensors?

1

As discussed in the meeting today, it's not clear that the current support for integer precision tensors is particularly useful. I noticed this when working to add unit test fixtures...

Enable building with fpm

5

Closes #77. This PR adds fpm support. https://fpm.fortran-lang.org/ I am currently building with ``` fpm build --cxx-flag "-cpp -std=c++17 -I${Torch_DIR}/include -I${Torch_DIR}/include/torch/csrc/api/include -L${Torch_DIR}/lib" --verbose ``` I get the following error but...

### Discussed in https://github.com/Cambridge-ICCS/FTorch/discussions/419 Originally posted by **rgknox** August 5, 2025 Hello. I've generated a trained neural network model in pytorch, and loaded it with Ftorch with the intention of...

Note: Found during this that one cannot use `tensor_a = tensor_b(requires_grad=.true.)` when `tensor_a` is only defined as a `torch_tensor`. This is because the `required_grad` property gets copied across, and then...