Crate syntaxdot_transformers

Transformer models (Vaswani et al., 2017)

This crate implements various transformer models, provided through the models module. The implementations are more restricted than e.g. their Huggingface counterparts, focusing only on the parts necessary for sequence labeling.

Modules

• activations
  Activation functions
• layers
  Basic neural network modules.
• loss
• models
  Transformer models.
• module
• scalar_weighting
  Scalar weighting of transformer layers.

Enums

• TransformerError
  Transformer errors.