MemoryBank<T>
Models & Types · AiDotNet.SelfSupervisedLearning
FIFO memory queue for storing embeddings in contrastive learning.
Implements: IMemoryBank<T>
For Beginners
A memory bank is a queue that stores embeddings from previous batches. This provides a large pool of negative samples for contrastive learning without requiring huge batch sizes.
How It Works
How it works:
Example usage:
Constructors
| Constructor | Summary |
|---|---|
MemoryBank(int capacity, int embeddingDim, int? seed) |
Initializes a new instance of the MemoryBank class. |
Properties
| Property | Summary |
|---|---|
bool IsFull { get; } |
|
int Capacity { get; } |
|
int CurrentSize { get; } |
|
int EmbeddingDimension { get; } |
Methods
| Method | Summary |
|---|---|
Tensor<T> GetAll() |
|
Tensor<T> GetAt(int index) |
Gets the embedding at a specific index. |
Tensor<T> Sample(int count) |
|
void Clear() |
|
void Enqueue(Tensor<T> embeddings) |
|
void SetAt(int index, Tensor<T> embedding) |
Sets the embedding at a specific index. |
void UpdateWithMomentum(int[] indices, Tensor<T> newEmbeddings, double momentum) |