please report a bug to PyTorch. We don‘t have an op for aten

.pth转pt遇到这个问题：

RuntimeError: hasSpecialCase INTERNAL ASSERT FAILED at /opt/conda/conda-bld/pytorch_1579027003190/work/torch/csrc/jit/passes/alias_analysis.cpp:300, please report a bug to PyTorch. We don’t have an op for aten

原代码如下：

    model = Model(opt)
    
    model = torch.nn.DataParallel(model).to(device)
    state_dict = torch.load(opt.saved_model)
    model.load_state_dict(state_dict, strict=True)

    for p in model.parameters():
        p.requires_grad = False
    model.eval()
    model = model.cuda()
    #print(model)
    
    example = torch.rand(1, 1, 48, 320).to(device)
    traced_script_module = torch.jit.trace(model, example)
    print("---------")
    print(traced_script_module)
    traced_script_module.save("./tps.pt")

将以下代码

traced_script_module = torch.jit.trace(model, example)

改为

traced_script_module = torch.jit.trace(model.module, example)

即可成功转为pt，不过记得要用上面第三行的 torch.nn.DataParallel。