ONNX LSTM-LM Export Fails

[abrarmatin]

I want to export a protobuf of the LSTM-LM in ONNX, but I'm failing with the error:
"ValueError: Auto nesting doesn't know how to process an input object of type float. Accepted types: Tensors, or lists/tuples of them"

I modeled my export function after word_language model:

def export_onnx(path, batch_size, seq_len):
print('The model is also exported in ONNX format at {}'.
format(os.path.realpath(args.onnx_export)))
model.eval()
dummy_input = torch.LongTensor(seq_len * batch_size).zero().view(-1, batch_size).to(device)
hidden = model.init_hidden(batch_size)
torch.onnx.export(model, (dummy_input, hidden), path)_

Searching online reveals a similar issue that suggests the bug is fixed with the newest onnx update: OpenNMT/OpenNMT-py#638

Is it possible to get a protobuf of the AWD-LSTM LM?