How to save torch tensors in Python that can be loaded in C++

[swahtz]

In Python:

import torch
from torch import nn

class TensorContainer(nn.Module):
    def __init__(self, tensor_dict):
        super().__init__()
        for key,value in tensor_dict.items():
            setattr(self, key, value)

prior = torch.ones(32145, 4)
tensor_dict = {'prior': prior}
tensors = TensorContainer(tensor_dict)
tensors = torch.jit.script(tensors)
tensors.save('prior.pth') # I believe this is the torch.jit.save() thing that your error message is asking for.

In C++

#include "torch/csrc/jit/import.h"
#include "torch/script.h"

void main(){
    torch::jit::script::Module tensors = torch::jit::load("prior.pth");
    torch::Tensor prior = tensors.get_attribute("prior").toTensor();
}

Originally posted by @fwilliams