bulkinferrer.md

The BulkInferrer TFX Pipeline Component

The BulkInferrer TFX component performs batch inference on unlabeled data. The generated InferenceResult(tensorflow_serving.apis.prediction_log_pb2.PredictionLog) contains the original features and the prediction results.

BulkInferrer consumes:

• A trained model in SavedModel format.

• Unlabelled tf.Examples that contain features.

• (Optional) Validation result from Evaluator component.

BulkInferrer emits:

• InferenceResult

Using the BulkInferrer Component

A BulkInferrer TFX component is used to perform batch inference on unlabeled tf.Examples. It is typically deployed after an Evaluator component to perform inference with a validated model, or after a Trainer component to directly perform inference on exported model.

It currently performs in-memory model inference and remote inference. Remote inference requires the model to be hosted on Cloud AI Platform.

Typical code looks like this:

bulk_inferrer = BulkInferrer(
    examples=examples_gen.outputs['examples'],
    model=trainer.outputs['model'],
    model_blessing=evaluator.outputs['blessing'],
    data_spec=bulk_inferrer_pb2.DataSpec(),
    model_spec=bulk_inferrer_pb2.ModelSpec()
)

More details are available in the BulkInferrer API reference.