AFTLanguageModel class

Decoder-only language model built from AFT blocks. Analogous to TransformerLM but with attention-free self-attention.

Inheritance

Constructors

AFTLanguageModel({required int vocabSize, required int embedDim, required int numLayers, required int maxLen, int? ffnDim, double dropoutP = 0.0, Device device = Device.CPU, int seed = 0})

blocks → List<AFTBlock>: final
embedDim → int: final
finalLn → LayerNorm: final
hashCode → int: The hash code for this object.
no setterinherited
head → Linear: final
maxLen → int: final
numLayers → int: final
posEnc → SinusoidalPositionalEncoding: final
runtimeType → Type: A representation of the runtime type of the object.
no setterinherited
tokenEmb → Embedding: final
training ↔ bool: Whether this module is in training mode. Layers that behave differently between training and inference (e.g. Dropout) read this flag in their call method. Defaults to training mode.
getter/setter pairinherited
vocabSize → int: final

call(Tensor tokens) → Tensor: Forward pass. tokens is 1D [seqLen] — returns logits [seqLen, vocabSize]. No batched 2D path yet (AFT attention module is 2D-only in this implementation).
eval() → void: Put this module (and any registered submodules) into evaluation mode.
inherited
noSuchMethod(Invocation invocation) → dynamic: Invoked when a nonexistent method or property is accessed.
inherited
parameters() → List<Tensor>: Trainable tensors owned by this module (and its submodules).
override
submodules() → List<Module>: Submodules owned by this module. Subclasses that compose other modules should override this so train() / eval() propagate. Default: empty.
override
toString() → String: A string representation of this object.
inherited
train() → void: Put this module (and any registered submodules) into training mode.
inherited
zeroGrad() → void: Zero every parameter's gradient. Safe to call before each backward.
inherited