how to write "platform_config_file" for TensorFlow serving

Question

I want to deploy a model with tensorflow serving. In the model, there are some ops that can't place into gpu. While deploy my model, I should config the tensorflow serving config file platform_config_file. The most important is allow_soft_placement. How to write it?

platform_configs: {
    key: "tensorflow",
    value:  {
        source_adapter_config: {
            type_url: "type.googleapis.com/tensorflow.serving.SavedModelBundleSourceAdapterConfig"
            value: 

And, it use any.proto, the value must be bytes.How to write this. thanks very much.