Pusher TFX 流水线组件
使用集合让一切井井有条
根据您的偏好保存内容并对其进行分类。
Pusher 组件用于在模型训练或再训练期间将经过验证的模型推送到部署目标。在部署之前,Pusher 依靠源于其他验证组件的一个或多个推荐来决定是否推送模型。
Pusher 组件使用 SavedModel 格式的训练模型,并生成相同的 SavedModel 以及版本控制元数据。
使用 Pusher 组件
Pusher 流水线组件通常非常易于部署,而且几乎不需要自定义,因为所有工作均由 Pusher TFX 组件完成。典型代码如下所示:
from tfx import components
...
pusher = components.Pusher(
model=trainer.outputs['model'],
model_blessing=evaluator.outputs['blessing'],
infra_blessing=infra_validator.outputs['blessing'],
push_destination=pusher_pb2.PushDestination(
filesystem=pusher_pb2.PushDestination.Filesystem(
base_directory=serving_model_dir)
)
)
如未另行说明,那么本页面中的内容已根据知识共享署名 4.0 许可获得了许可,并且代码示例已根据 Apache 2.0 许可获得了许可。有关详情,请参阅 Google 开发者网站政策。Java 是 Oracle 和/或其关联公司的注册商标。
最后更新时间 (UTC):2020-11-13。
[null,null,["最后更新时间 (UTC):2020-11-13。"],[],[],null,["# The Pusher TFX Pipeline Component\n\n\u003cbr /\u003e\n\nThe Pusher component is used to push a validated model to a\n[deployment target](/tfx/guide#deployment_targets) during model training or\nre-training. Before the deployment, Pusher relies on one or more blessings from\nother validation components to decide whether to push the model or not.\n\n- [Evaluator](/tfx/guide/evaluator) blesses the model if the new trained model is \"good enough\" to be pushed to production.\n- (Optional but recommended) [InfraValidator](/tfx/guide/infra_validator) blesses the model if the model is mechanically servable in a production environment.\n\nA Pusher component consumes a trained model in [SavedModel](/guide/saved_model)\nformat, and produces the same SavedModel, along with versioning metadata.\n\nUsing the Pusher Component\n--------------------------\n\nA Pusher pipeline component is typically very easy to deploy and requires little\ncustomization, since all of the work is done by the Pusher TFX component.\nTypical code looks like this: \n\n pusher = Pusher(\n model=trainer.outputs['model'],\n model_blessing=evaluator.outputs['blessing'],\n infra_blessing=infra_validator.outputs['blessing'],\n push_destination=tfx.proto.PushDestination(\n filesystem=tfx.proto.PushDestination.Filesystem(\n base_directory=serving_model_dir)\n )\n )\n\n### Pushing a model produced from InfraValidator.\n\n(From version 0.30.0)\n\nInfraValidator can also produce `InfraBlessing` artifact containing a\n[model with warmup](/tfx/guide/infra_validator#producing_a_savedmodel_with_warmup), and\nPusher can push it just like a `Model` artifact. \n\n infra_validator = InfraValidator(\n ...,\n # make_warmup=True will produce a model with warmup requests in its\n # 'blessing' output.\n request_spec=tfx.proto.RequestSpec(..., make_warmup=True)\n )\n\n pusher = Pusher(\n # Push model from 'infra_blessing' input.\n infra_blessing=infra_validator.outputs['blessing'],\n push_destination=tfx.proto.PushDestination(...)\n )\n\nMore details are available in the\n[Pusher API reference](https://www.tensorflow.org/tfx/api_docs/python/tfx/v1/components/Pusher)."]]