Expand description
builtin ai layers and value types
Modules§
- math
- layers representing important mathematical operations
- reinforcement
- layers relating to reinforcement learning
- soft
- layers relating to cross entropy or softmax
- structural
- layers like cat or swap dims that change the organization or structure of tensors
Structs§
- Apply
- applies a closure to the input// TODO more closure layers maybe
- Autoregression
- autoregressive inference
- Duplicate
- module for cloning things
- Map
- wraps to apply to every element of a vector
- Residual
- layer to add an optional residual connection
- Sequential
- wrapper for applying ai modules sequentially
- SetType
- fixes the output type of a layer for a particular input type.
- Squeeze
Layer - layer for removing dimensions with size 1
- Zip
- wraps to apply each function
Enums§
- Alignment
- alignment
- OnMismatch
- shape mismatch handling
- Reduction
Mode - reduction mode
Functions§
- apply
- creates an operation that applies the closure