SupervisedFineTuning<T, TInput, TOutput>

Models & Types ยท AiDotNet.FineTuning

Implements Supervised Fine-Tuning (SFT) - the foundational fine-tuning method.

Inheritance: FineTuningBase<T, TInput, TOutput> โ†’ SupervisedFineTuning<T, TInput, TOutput>

For Beginners

SFT is like teaching by example. You show the model many examples of correct input-output pairs, and it learns to produce similar outputs. For instance, you might train a model on high-quality question-answer pairs to make it better at answering questions.

How It Works

SFT trains a model on labeled input-output pairs using standard supervised learning. This is the most straightforward fine-tuning approach and serves as the foundation for more advanced methods like RLHF or DPO.

Use Cases:

Constructors

Constructor Summary
SupervisedFineTuning(FineTuningOptions<T> options) Initializes a new instance of SFT with default loss function.
SupervisedFineTuning(FineTuningOptions<T> options, ILossFunction<T> lossFunction) Initializes a new instance of SFT with a custom loss function.

Properties

Property Summary
FineTuningCategory Category { get; }
bool RequiresReferenceModel { get; }
bool RequiresRewardModel { get; }
string MethodName { get; }

Methods

Method Summary
Task<FineTuningMetrics<T>> EvaluateAsync(IFullModel<T, TInput, TOutput> model, FineTuningData<T, TInput, TOutput> evaluationData, CancellationToken cancellationToken)
Task<IFullModel<T, TInput, TOutput>> FineTuneAsync(IFullModel<T, TInput, TOutput> baseModel, FineTuningData<T, TInput, TOutput> trainingData, CancellationToken cancellationToken)