SelfSupervisedLearningFineTuningPipeline<T>
Models & Types · AiDotNet.SelfSupervisedLearning
Pipeline for fine-tuning SSL pretrained encoders on downstream tasks.
For Beginners
After SSL pretraining, you typically want to fine-tune the encoder on a specific task with labeled data. This pipeline handles the fine-tuning process with proper learning rate schedules and evaluation.
How It Works
Fine-tuning strategies:
Constructors
| Constructor | Summary |
|---|---|
SelfSupervisedLearningFineTuningPipeline(INeuralNetwork<T> encoder, int encoderOutputDim, int numClasses) |
Initializes a new fine-tuning pipeline. |
Methods
| Method | Summary |
|---|---|
SelfSupervisedLearningFineTuningPipeline<T> WithConfig(Action<FineTuningConfig> configure) |
Configures fine-tuning parameters. |
SelfSupervisedLearningFineTuningPipeline<T> WithStrategy(FineTuningStrategy strategy) |
Sets the fine-tuning strategy. |
double Evaluate(Tensor<T> testData, int[] testLabels) |
Evaluates the model on test data. |