solved onnx inference tensor to numpy conversion

[saifulislam79]

ONNX Inference

The model supports ONNX inference. PyTorch tensors are automatically converted to numpy arrays for compatibility:

# Tensors are automatically converted to numpy arrays
if isinstance(codes, torch.Tensor):
    codes = codes.detach().cpu().numpy()

This ensures seamless integration between PyTorch and ONNX runtime.

[harryjulian]

happy to merge if you remove the changes to the README.