EagerInitializationStrategy<T>
Models & Types · AiDotNet.Initialization
Eager initialization strategy that initializes weights immediately on construction.
Inheritance: InitializationStrategyBase<T> → EagerInitializationStrategy<T>
How It Works
This is the traditional initialization approach where weights are allocated and initialized during layer construction. This ensures all weights are ready before any training or inference begins.
Constructors
| Constructor | Summary |
|---|---|
EagerInitializationStrategy() |
Initializes a new instance using the framework’s default non-deterministic RNG. |
EagerInitializationStrategy(Random rng) |
Initializes a new instance using the supplied Random for reproducible weight initialization. |
Properties
| Property | Summary |
|---|---|
bool IsLazy { get; } |
|
bool LoadFromExternal { get; } |
Methods
| Method | Summary |
|---|---|
IInitializationStrategy<T> WithSeededRandom(Random rng) |
|
void InitializeBiases(Tensor<T> biases) |
|
void InitializeWeights(Tensor<T> weights, int inputSize, int outputSize) |