beyondml.tflow package

Subpackages

• beyondml.tflow.layers package
  • Submodules
  • beyondml.tflow.layers.FilterLayer module
    • FilterLayer
      • FilterLayer.call()
      • FilterLayer.from_config()
      • FilterLayer.get_config()
      • FilterLayer.turn_off()
      • FilterLayer.turn_on()
  • beyondml.tflow.layers.MaskedConv2D module
    • MaskedConv2D
      • MaskedConv2D.build()
      • MaskedConv2D.call()
      • MaskedConv2D.from_config()
      • MaskedConv2D.get_config()
      • MaskedConv2D.kernel_size
      • MaskedConv2D.set_masks()
  • beyondml.tflow.layers.MaskedConv3D module
    • MaskedConv3D
      • MaskedConv3D.build()
      • MaskedConv3D.call()
      • MaskedConv3D.from_config()
      • MaskedConv3D.get_config()
      • MaskedConv3D.kernel_size
      • MaskedConv3D.set_masks()
  • beyondml.tflow.layers.MaskedDense module
    • MaskedDense
      • MaskedDense.build()
      • MaskedDense.call()
      • MaskedDense.from_config()
      • MaskedDense.get_config()
      • MaskedDense.set_masks()
  • beyondml.tflow.layers.MultiConv2D module
    • MultiConv2D
      • MultiConv2D.build()
      • MultiConv2D.call()
      • MultiConv2D.from_config()
      • MultiConv2D.get_config()
      • MultiConv2D.kernel_size
  • beyondml.tflow.layers.MultiConv3D module
    • MultiConv3D
      • MultiConv3D.build()
      • MultiConv3D.call()
      • MultiConv3D.from_config()
      • MultiConv3D.get_config()
      • MultiConv3D.kernel_size
  • beyondml.tflow.layers.MultiDense module
    • MultiDense
      • MultiDense.build()
      • MultiDense.call()
      • MultiDense.from_config()
      • MultiDense.get_config()
  • beyondml.tflow.layers.MultiMaskedConv2D module
    • MultiMaskedConv2D
      • MultiMaskedConv2D.build()
      • MultiMaskedConv2D.call()
      • MultiMaskedConv2D.from_config()
      • MultiMaskedConv2D.get_config()
      • MultiMaskedConv2D.kernel_size
      • MultiMaskedConv2D.set_masks()
  • beyondml.tflow.layers.MultiMaskedConv3D module
    • MultiMaskedConv3D
      • MultiMaskedConv3D.build()
      • MultiMaskedConv3D.call()
      • MultiMaskedConv3D.from_config()
      • MultiMaskedConv3D.get_config()
      • MultiMaskedConv3D.kernel_size
      • MultiMaskedConv3D.set_masks()
  • beyondml.tflow.layers.MultiMaskedDense module
    • MultiMaskedDense
      • MultiMaskedDense.build()
      • MultiMaskedDense.call()
      • MultiMaskedDense.from_config()
      • MultiMaskedDense.get_config()
      • MultiMaskedDense.set_masks()
  • beyondml.tflow.layers.MultiMaxPool2D module
    • MultiMaxPool2D
      • MultiMaxPool2D.call()
      • MultiMaxPool2D.from_config()
      • MultiMaxPool2D.get_config()
  • beyondml.tflow.layers.MultiMaxPool3D module
    • MultiMaxPool3D
      • MultiMaxPool3D.call()
      • MultiMaxPool3D.from_config()
      • MultiMaxPool3D.get_config()
  • beyondml.tflow.layers.MultitaskNormalization module
    • MultitaskNormalization
      • MultitaskNormalization.build()
      • MultitaskNormalization.call()
      • MultitaskNormalization.from_config()
      • MultitaskNormalization.get_config()
  • beyondml.tflow.layers.SelectorLayer module
    • SelectorLayer
      • SelectorLayer.call()
      • SelectorLayer.from_config()
      • SelectorLayer.get_config()
      • SelectorLayer.sel_index
  • beyondml.tflow.layers.SparseConv2D module
    • SparseConv2D
      • SparseConv2D.build()
      • SparseConv2D.call()
      • SparseConv2D.from_config()
      • SparseConv2D.from_layer()
      • SparseConv2D.get_config()
  • beyondml.tflow.layers.SparseConv3D module
    • SparseConv3D
      • SparseConv3D.build()
      • SparseConv3D.call()
      • SparseConv3D.from_config()
      • SparseConv3D.from_layer()
      • SparseConv3D.get_config()
  • beyondml.tflow.layers.SparseDense module
    • SparseDense
      • SparseDense.build()
      • SparseDense.call()
      • SparseDense.from_config()
      • SparseDense.from_layer()
      • SparseDense.get_config()
  • beyondml.tflow.layers.SparseMultiConv2D module
    • SparseMultiConv2D
      • SparseMultiConv2D.build()
      • SparseMultiConv2D.call()
      • SparseMultiConv2D.from_config()
      • SparseMultiConv2D.from_layer()
      • SparseMultiConv2D.get_config()
  • beyondml.tflow.layers.SparseMultiConv3D module
    • SparseMultiConv3D
      • SparseMultiConv3D.build()
      • SparseMultiConv3D.call()
      • SparseMultiConv3D.from_config()
      • SparseMultiConv3D.from_layer()
      • SparseMultiConv3D.get_config()
  • beyondml.tflow.layers.SparseMultiDense module
    • SparseMultiDense
      • SparseMultiDense.build()
      • SparseMultiDense.call()
      • SparseMultiDense.from_config()
      • SparseMultiDense.from_layer()
      • SparseMultiDense.get_config()
  • beyondml.tflow.layers.SumLayer module
    • SumLayer
      • SumLayer.call()
      • SumLayer.from_config()
      • SumLayer.get_config()
  • Module contents
• beyondml.tflow.utils package
  • Submodules
  • beyondml.tflow.utils.transformer module
    • build_token_position_embedding_block()
    • build_transformer_block()
  • beyondml.tflow.utils.utils module
    • ActiveSparsification
      • ActiveSparsification.on_epoch_end()
      • ActiveSparsification.on_train_begin()
    • add_layer_masks()
    • get_custom_objects()
    • get_task_masking_gradients()
    • mask_model()
    • mask_task_weights()
    • quantize_model()
    • remove_layer_masks()
    • train_model()
    • train_model_iteratively()
  • Module contents

Module contents

## TensorFlow compatibility for building MANN models.

The beyondml.tflow package contains two subpackages, beyondml.tflow.layers and beyondml.tflow.utils, which contain the functionality to create and train MANN layers within TensorFlow. For individuals who are familiar with the former name of this package, mann, backwards compatibility can be achieved (assuming only TensorFlow support is needed), by replacing the following line of code:

>>> import mann

with the following line:

>>> import beyondml.tflow as mann

in all existing scripts.

Within the layers package, there is current functionality for the the following layers: - beyondml.tflow.layers.FilterLayer - beyondml.tflow.layers.MaskedConv2D - beyondml.tflow.layers.MaskedDense - beyondml.tflow.layers.MultiConv2D - beyondml.tflow.layers.MultiDense - beyondml.tflow.layers.MultiMaskedConv2D - beyondml.tflow.layers.MultiMaskedDense - beyondml.tflow.layers.MultiMaxPool2D - beyondml.tflow.layers.SelectorLayer - beyondml.tflow.layers.SumLayer - beyondml.tflow.layers.SparseDense - beyondml.tflow.layers.SparseConv - beyondml.tflow.layers.SparseMultiDense - beyondml.tflow.layers.SparseMultiConv

Note that with any of the sparse layers (such as the `SparseDense` layer), any model which utilizes these layers will not be loadable using the traditional `load_model` functions available in TensorFlow. Instead, the model should be saved using either joblib or pickle.

Within the utils package, there are the current functions and classes: - ActiveSparsification - build_transformer_block - build_token_position_embedding_block - get_custom_objects - mask_model - remove_layer_masks - add_layer_masks - quantize_model - get_task_masking_gradients - mask_task_weights - train_model_iteratively