AiDotNet API Reference

Welcome to the AiDotNet API Reference documentation. This section provides complete API documentation for all public classes, interfaces, methods, and properties.

Core Namespaces

Namespace	Description
AiDotNet	Core builder and result types (AiModelBuilder, AiModelResult)
AiDotNet.Configuration	Configuration options and settings
AiDotNet.NeuralNetworks	100+ neural network architectures
AiDotNet.Classification	28+ classification algorithms
AiDotNet.Regression	41+ regression algorithms
AiDotNet.Clustering	20+ clustering algorithms
AiDotNet.ComputerVision	50+ vision models (YOLO, DETR, SAM)
AiDotNet.Audio	90+ audio models (Whisper, TTS)
AiDotNet.ReinforcementLearning	80+ RL agents (DQN, PPO, SAC)
AiDotNet.Diffusion	20+ diffusion models
AiDotNet.LoRA	37+ LoRA adapters (QLoRA, DoRA)
AiDotNet.RAG	50+ RAG components
AiDotNet.DistributedTraining	DDP, FSDP, ZeRO strategies
AiDotNet.AutoML	Automatic model selection
AiDotNet.Serving	Production model serving
AiDotNet.Tensors	Tensor operations and linear algebra
AiDotNet.Tokenization	Text tokenization (BPE, WordPiece)

Quick Start

using AiDotNet;

// Build and train a model using the facade pattern
var result = await new AiModelBuilder<double, double[], double>()
    .ConfigureModel(new NeuralNetwork<double>(inputSize: 10, hiddenSize: 64, outputSize: 2))
    .ConfigureOptimizer(new AdamOptimizer<double>())
    .ConfigurePreprocessing()
    .BuildAsync(features, labels);

// Make predictions using the result directly
var prediction = result.Predict(newSample);

Entry Points

AiModelBuilder

The AiModelBuilder<T, TInput, TOutput> class is your primary entry point for building and training models. It uses a fluent builder pattern for configuration.

Key methods:

ConfigureModel() - Set the model architecture
ConfigureOptimizer() - Set the training optimizer
ConfigurePreprocessing() - Configure data preprocessing
ConfigureAutoML() - Enable automatic model selection
ConfigureHuggingFace() - Load HuggingFace models
ConfigureDistributedTraining() - Enable multi-GPU training
BuildAsync() - Build and train the model

AiModelResult

The AiModelResult<T, TInput, TOutput> class wraps your trained model and provides inference capabilities.

Key properties and methods:

Predict() - Make predictions on new data
Model - Access the underlying trained model
Metrics - Training and validation metrics
Save() / Load() - Model persistence