Question

我正在尝试在https://github.com/google/sentencepiece/tree/master/tensorflow的模型中使用tf-sentencepiece操作

构建模型并获取带有变量和资产的saved_model.pb文件没有问题。但是，如果我尝试将docker映像用于tensorflow / serving，它会显示

Loading servable: {name: model version: 1} failed: 
Not found: Op type not registered 'SentencepieceEncodeSparse' in binary running on 0ccbcd3998d1. 
Make sure the Op and Kernel are registered in the binary running in this process. 
Note that if you are loading a saved graph which used ops from tf.contrib, accessing 
(e.g.) `tf.contrib.resampler` should be done before importing the graph, 
as contrib ops are lazily registered when the module is first accessed.

我不熟悉如何手动构建任何东西，希望我可以做很多事情而无需做任何改动。

Answer 1

一种方法是：

拉出docker开发映像

$ docker pull tensorflow / serving：最新开发
在容器中，更改代码

$ docker run -it tensorflow / serving：latest-devel

修改代码以添加op依赖项here。

在容器中，构建TensorFlow Serving

容器：$ tensorflow_serving / model_servers：tensorflow_model_server && cp bazel-bin / tensorflow_serving / model_servers / tensorflow_model_server / usr / local / bin /
使用exit命令退出容器
查找容器ID：

$ docker ps
使用该容器ID提交开发映像：

$ docker commit $ USER / tf-serving-devel-custom-op
现在使用开发容器作为源构建服务容器

$ mkdir / tmp / tfserving

$ cd / tmp / tfserving

$ git clone https://github.com/tensorflow/serving。

$ docker build -t $ USER / tensorflow-serving --build-arg TF_SERVING_BUILD_IMAGE = $ USER / tf-serving-devel-custom-op -f tensorflow_serving / tools / docker / Dockerfile。
您现在可以使用$ USER / tensorflow-serving在Docker instructions

有自定义操作时如何使用docker image tensorflow / serving服务tensorflow模型？

1 个答案: