Tensorflow 服务:“没有要保存/写入的资产"导出模型时 [英] Tensorflow serving: "No assets to save/writes" when exporting models

问题描述

最近我正在尝试使用 tensorflow 服务部署深度学习服务.但是在导出模型时我得到了以下信息:

Recently I am trying to deploy deep learning services using tensorflow serving. But I got the following infos when exporting my model:

  INFO:tensorflow: No assets to save
  INFO:tensorflow: No assets to write
  INFO:tensorflow: SavedModel written to: b'./models/1/saved_model.pb'

我真的不明白这里发生了什么.没有要保存/写入的资产"是什么意思?一切顺利吗?顺便说一句，通过运行官方示例 Serving a tensorflow model，我得到了相同的信息.

I don't really understand what happening here. What does "No assets to save/write" mean? Is everthing goes well? btw，by running the official example Serving a tensorflow model, I got the same infos.

推荐答案

资产是指您的模型所需的任何外部文件.常见的例子是词汇表和嵌入矩阵.如果您的模型没有任何外部依赖项，则会出现这些消息.请注意，它们记录在信息"级别.

Assets mean any external files that are needed for you model. Common examples are vocabularies and embedding matrices. If your model does not have any external dependencies, these messages are expected. Note that they are logged at "INFO" level.

这篇关于Tensorflow 服务:“没有要保存/写入的资产"导出模型时的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！