转换RIFE模型到onnx再转换到tensorfow失败

一、转换代码如下：
onnx_path = 'flownet.onnx'
dummy_input = torch.randn(1, 6, 256, 448).to(device)
torch.onnx.export(self.flownet,
dummy_input,
onnx_path,
export_params=True,
opset_version=16,
do_constant_folding=True,
input_names = ['input'],
output_names = ['output'],
)
print(f'Model has been exported to {onnx_path}')

   onnx_model = onnx.load(onnx_path)
   tf_rep = prepare(onnx_model)
   tf_rep.export_graph("flownet.pb")
   ## 加载 SavedModel
   model = tf.saved_model.load("flownet.pb")

二、报错如下：
BackendIsNotSupposedToImplementIt: LeakyRelu version 16 is not implemented.

如果opset_version设置11则提示grid_sample要opset_version 16

或者请问下能不能提供Tf或者tflite的预训练模型？

Start by reproducing the shown torch.onnx.export call with opset 16, then run onnx_tf.prepare on flownet.onnx and inspect the LeakyRelu backend error alongside the grid_sample requirement. Done would mean establishing a compatible conversion path or clearly determining whether a TensorFlow or TFLite pretrained model can be provided.