"Error: Expected tensor for argument #1 'indices' to have scalar type Long; but got CPUFloatType instead"

[JinTonic101]

Hi @vvmnnnkv

So as discussed before, the lib only supports float tensors and it can only converts back to JS as Uint8 array atm. It could be great to expose more types/functions and make it more generic.

In C++ I've managed to fix the issue by changing the input tensor type to 'kLong':
torch::jit::IValue input = torch::from_blob(data, shape, torch::TensorOptions().dtype(torch::kLong))

Thank you in advance and let me know if I can help in any way !