Module fontai.prediction

This submodule contains Tensorflow-related classes that preprocess and format model inputs and define custom Keras models and callbacks. Arbitrary sequential Keras models can be instantiated from configurations, but user-defined custom models can be used as well by defining them in the fontai.prediction.models module.

Expand source code
"""This submodule contains Tensorflow-related classes that preprocess and format model inputs and define custom Keras models and callbacks. Arbitrary sequential Keras models can be instantiated from configurations, but user-defined custom models can be used as well by defining them in the `models` module.
  
"""

Sub-modules

fontai.prediction.callbacks

This module contains custom Tensorflow callbacks

fontai.prediction.custom_filters

This module contains filtering functions for Tensorflow dataset operations that are applied to model inputs right after being deserialised to be used …

fontai.prediction.custom_mappers

This module contains mapper functions to be applied to Tensorflow examples right after being deserialised to be used at training time; at the moment …

fontai.prediction.input_processing

This module contains input preprocessing logic that happens right before data is ingested by the model to be trained.

fontai.prediction.models