ModelMode.md

tfra.dynamic_embedding.ModelMode

View source on GitHub

Class ModelMode

The global config of model modes.

TrainableWrapper.read_value is not thread-safe that causes threads competition and Out-Of-Bound exception in concurrent serving scenario.

To resolve this, we define the ModelMode APIs to instruct the TrainableWrapper to build a different thread-safe sub-graph for 'TrainableWrapper.read_value' on inference mode.

NOTE These APIs should be called before any graph are built.

The following standard modes are defined:

• TRAIN: training/fitting mode.
• INFERENCE: prediction/inference mode.

Class Members

• CURRENT_SETTING = 'train'
• INFERENCE = 'inference'
• TRAIN = 'train'