TFX 运行多个培训师同时

Question

我是 tfx 的新手，正在学习组装管道。我已经在 GCP 上的 Kubeflow 中成功构建了一个管道。我想知道如何使用相同的 csvexamplegen、转换和 schemagen 运行多个训练器在管道中生成不同的 outputs/pushers。有没有人这样做过？请提前告知并感谢。

  def create_pipeline(
    pipeline_name: Text,
    pipeline_root: Text,
    data_path: Text,
    preprocessing_fn: Text,
    run_fn: Text,
    train_args: trainer_pb2.TrainArgs,
    eval_args: trainer_pb2.EvalArgs,
    eval_accuracy_threshold: float,
    serving_model_dir: Text,
    metadata_connection_config: Optional[
        metadata_store_pb2.ConnectionConfig] = None
) -> pipeline.Pipeline:


      trainer_args = {
      'run_fn': run_fn,
      'transformed_examples': transform.outputs['transformed_examples'],
      'schema': schema_gen.outputs['schema'],
      'transform_graph': transform.outputs['transform_graph'],
      'train_args': train_args,
      'eval_args': eval_args,
      'custom_executor_spec':
          executor_spec.ExecutorClassSpec(trainer_executor.GenericExecutor)}

      trainer = Trainer(**trainer_args)
      components.append(trainer)

return pipeline.Pipeline(
      pipeline_name=pipeline_name,
      pipeline_root=pipeline_root,
      components=components,
      enable_cache=True,
      metadata_connection_config=metadata_connection_config,
      beam_pipeline_args=beam_pipeline_args,
  )

Answer 1

为了让 Kubeflow 拾取的培训师独一无二，我必须输入 instance_name 来定义培训师。

trainer2 = Trainer(
      run_fn=run_fn2,
      examples=transform.outputs['transformed_examples'],
      schema=schema_gen.outputs['schema'],
      transform_graph= transform.outputs['transform_graph'],
      train_args= train_args2,
      eval_args= eval_args2,
      custom_executor_spec= executor_spec.ExecutorClassSpec(trainer_executor.GenericExecutor),
      instance_name='trainer2'
  )

TFX 运行多个培训师同时

TFX running multiple trainers simultaneously

python

tensorflow2.0

tensorflow-serving

tfx

kubeflow-pipelines

TFX 运行 多个培训师同时

TFX running multiple trainers simultaneously

python

tensorflow2.0

tensorflow-serving

tfx

kubeflow-pipelines

TFX 运行多个培训师同时