Models

764 public types in this namespace, organized by kind.

Models & Types (143)

Type Summary
AiDotNetLicenseKey Represents a license key for AiDotNet model encryption and online validation.
AiModelResult<T, TInput, TOutput> Partial class containing Test-Time Augmentation (TTA) prediction methods.
AlignmentEvaluationData<T> Contains test cases for evaluating AI alignment.
AlignmentFeedbackData<T> Contains human feedback data for AI alignment.
AlignmentMetrics<T> Contains metrics for evaluating AI alignment with human values.
AutoMLRunSummary Represents a redacted (safe-to-share) summary of an AutoML run.
AutoMLTrialSummary Represents a redacted (safe-to-share) summary of a single AutoML trial.
BootstrapResult<T> Represents the results of bootstrap validation for a machine learning model, containing R² metrics for training, validation, and test datasets.
CategoricalColumnStats Statistics for a categorical column.
CategoricalDistribution Represents a categorical (discrete choice) parameter distribution.
CertifiedAccuracyMetrics<T> Contains metrics for certified accuracy evaluation.
CertifiedPrediction<T> Represents a certified prediction with robustness guarantees.
CheckpointMetadata<T> Contains metadata about a checkpoint without loading the full checkpoint data.
Checkpoint<T, TInput, TOutput> Represents a saved checkpoint of model training state.
ChiSquareTestResult<T> Represents the results of a Chi-Square statistical test, which is used to determine whether there is a significant association between two categorical variables.
ClassificationConformalPredictionSet Represents a conformal prediction set for classification tasks.
ClientSelectionRequest Represents a request to select participating clients for a federated learning round.
CloneConstructorArgumentBinding Retains the declaring identity of constructor state, including private or shadowed members.
ClonePlan Describes how to reproduce one type: which members are configuration, and how each is carried.
ClonePlanEntry One configuration member and the manner in which a clone carries it.
ClusteringMetrics<T> Represents clustering quality metrics for evaluating the performance of clustering algorithms.
ComponentAccessorParameterSource<T> A lazily obtained component whose identity is registered before the component itself exists.
ComponentCollectionParameterSource<T> A COLLECTION of parameterized components exposed as one parameter surface, concatenated in enumeration order.
ContinuousDistribution Represents a continuous (real-valued) parameter distribution.
CrossValidationResult<T, TInput, TOutput> Aggregates results from all folds in a cross-validation procedure.
DataSetStats<T, TInput, TOutput> Represents a comprehensive collection of statistical measures and data for evaluating model performance on a dataset.
DatasetComparison<T> Comparison results between two dataset versions.
DatasetLineage Lineage information for a dataset showing its origin and transformations.
DatasetResult<T, TInput, TOutput> Represents detailed results and statistics for a specific dataset (training, validation, or test).
DatasetSnapshot A snapshot of a dataset at a point in time.
DatasetStatistics<T> Statistical summary of a dataset.
DatasetVersionInfo<T> Information about a dataset version.
DatasetVersion<T> Represents a versioned dataset with integrity verification.
DelegatingParameterSource<T> Adapts a component that already exposes count / read / write but does not implement IParameterSource.
DistributionFitResult<T> Represents the result of fitting a statistical distribution to a dataset, including the distribution type, goodness of fit measure, and estimated parameters.
DoubleArrayParameterSource<T> Exposes an explicitly-marked Double array as a write-through parameter source.
DoubleJaggedParameterSource<T> Exposes an explicitly-marked jagged Double array in stable row-major order.
DoubleScalarParameterSource<T> Exposes an explicitly-marked Double field as one write-through parameter.
DynamicShapeInfo Describes which dimensions of a model’s input/output shapes are dynamic (variable at runtime).
EpochHistory Represents the training history for a single epoch or iteration.
EvolutionEliteSummary One archived elite, described by identity and coordinates rather than by its contents.
EvolutionFailureSummary One retained failure from an evolution run, reduced to a code and a message.
EvolutionIslandSummary How one island of an evolution run ended: its coverage, its best entry, and how far it advanced.
EvolutionRunSummary A redacted, serializable account of one finished evolution run.
Experiment Represents a machine learning experiment that groups related training runs.
ExperimentInfo<T> Contains structured experiment tracking information from a trained model.
ExperimentRun<T> Represents a single training run within an experiment.
FTestResult<T> Represents the results of an F-test, which is used to compare the variances of two populations.
FederatedClientDataset<TInput, TOutput> Represents a single client’s local dataset for federated learning.
FederatedLearningMetadata Contains metadata and metrics about federated learning training progress and results.
FilterAction Represents a filtering action taken on the output.
FineTuningData<T, TInput, TOutput> Container for fine-tuning training and evaluation data.
FineTuningMetrics<T> Metrics for evaluating fine-tuning quality.
FitDetectorResult<T> Represents the result of a model fit detection analysis, which evaluates how well a model fits the data and provides recommendations for improvement.
FoldResult<T, TInput, TOutput> Represents the results of a single fold in cross-validation.
GradientModel<T> Default implementation of a gradient model.
HarmfulContentFinding Represents a specific harmful content finding.
HarmfulContentResult<T> Result of harmful content identification.
HyperparameterApplicationResult Contains the results of applying agent-recommended hyperparameters to a model’s options.
HyperparameterOptimizationResult<T> Contains the results of a hyperparameter optimization process.
HyperparameterSearchSpace Defines the search space for hyperparameter optimization.
HyperparameterTrial<T> Represents a single trial in hyperparameter optimization.
InferenceSequence<T, TInput, TOutput> Represents one independent, stateful inference sequence (e.g., one chat/generation stream).
InferenceSession<T, TInput, TOutput> Facade-friendly inference session that owns stateful inference internals.
IntegerDistribution Represents an integer parameter distribution.
InterventionEffect<T> Represents the effect of an intervention in a time series or sequential data, capturing the starting point, duration, and magnitude of the effect.
InterventionInfo Represents information about an intervention in a time series or sequential data, specifying when it started and how long it lasted.
JailbreakDetectionResult<T> Result of jailbreak attempt detection.
JailbreakIndicator Represents a specific jailbreak indicator.
KeyedMatrixCollectionParameterSource<T, TKey> A key-stable matrix dictionary exposed as one parameter source.
KeyedScalarCollectionParameterSource<T, TKey> A scalar dictionary exposed in canonical key order.
KeyedTensorCollectionParameterSource<T, TKey> A key-stable tensor dictionary exposed as one parameter source.
KeyedVectorCollectionParameterSource<T, TKey> A key-stable vector dictionary exposed as one parameter source.
KnowledgeGraphResult<T> Contains the results of knowledge graph processing, including trained embeddings, community structure, and link prediction evaluation.
LayerQuantizationInfo Contains quantization information for a specific layer.
LicenseValidationResult Contains the result of a license key validation attempt.
MannWhitneyUTestResult<T> Represents the results of a Mann-Whitney U test, which is a non-parametric statistical test used to determine whether two independent samples come from the same distribution.
MatrixCollectionParameterSource<T> An index-stable matrix collection exposed as one parameter source.
MatrixFieldParameterSource<T> A Matrix field exposed as a parameter surface, written through.
MetaAdaptationResult<T> Results from adapting a meta-learner to a single task.
MetaEvaluationResult<T> Results from evaluating a meta-learner across multiple tasks.
MetaTrainingResult<T> Results from a complete meta-training run with history tracking.
MetaTrainingStepResult<T> Results from a single meta-training step (one outer loop update).
ModelComparison<T> Comparison results between two model versions.
ModelEvaluationData<T, TInput, TOutput> Represents a comprehensive collection of evaluation data for a model across training, validation, and test datasets.
ModelEvaluationInput<T, TInput, TOutput> Represents the input data required for evaluating a machine learning model.
ModelLineage Lineage information for a model showing its origin and dependencies.
ModelMetadata<T> Represents metadata about a machine learning model, including its type, complexity, and additional descriptive information.
ModelRegistryInfo<T, TInput, TOutput> Contains structured model registry information from a trained model.
ModelSearchCriteria<T> Criteria for searching models in the registry.
ModelStateRegistry<T> The declared home for model state that is not a flat parameter vector.
ModelStatsInputs<T, TInput, TOutput> Represents a container for inputs needed to calculate various statistics and metrics for a model.
ModelVersionInfo<T> Information about a specific model version.
NASResultSummary Represents a redacted summary of Neural Architecture Search (NAS) results.
NestedKeyedScalarCollectionParameterSource<T, TOuterKey, TInnerKey> A two-level scalar dictionary exposed in canonical outer- and inner-key order.
NodeShape<T, TNode> Describes ONE node of a recursive structure, so the registry can walk the whole of it.
NormalizationParameters<T> Represents the parameters used for normalizing a single feature or target variable in a machine learning model.
NumericColumnStats<T> Statistics for a numeric column.
OFASubnetSummary Represents OnceForAll subnet configuration summary.
OptimizationInputData<T, TInput, TOutput> Represents the input data for optimization processes, including training, validation, and test datasets.
OptimizationIterationInfo<T> Represents information about a single iteration in an optimization process, including fitness and overfitting detection results.
OptimizationResult<T, TInput, TOutput> Represents the comprehensive results of an optimization process for a symbolic model, including the best solution found, performance metrics, feature selection results, and detailed statistics for different datasets.
OptimizationStepData<T, TInput, TOutput> Represents comprehensive data about a single step in an optimization process, including the current solution, selected features, data subsets, fitness scores, and evaluation results.
ParameterChunk<T> One concrete, ordered model-state chunk together with the manifest identity and semantic role that govern it.
ParameterComponentRegistry<T> Owns a model’s deterministic parameter manifest and derives count, flat values and restore from the same ordered component snapshot.
ParameterIndexRange Represents a contiguous parameter index range.
ParameterLayoutSnapshot A single deterministic snapshot consumed by count, vector, restore and checkpoint code.
ParameterSlotDescriptor An immutable, stable description of one model-state slot.
PermutationTestResult<T> Represents the results of a permutation test, which is a non-parametric statistical significance test that determines whether the observed difference between two groups is statistically significant.
QuantizationInfo Contains information about model quantization applied during or after training.
RecipeAdaptation One adjustment made to a paper’s recipe, and the rule that justifies it.
RedTeamingResults Contains results from red teaming a model.
RegisteredModel<T, TInput, TOutput> Represents a registered model in the model registry with its metadata and versioning information.
RegressionConformalInterval<TOutput> Represents a conformal prediction interval for regression-style outputs.
ResizableTensorFieldParameterSource<T> A tensor field whose one unresolved axis is learned from the first restore payload.
ResourceUsageStats Contains statistics about system resource usage during training.
RobustnessMetrics<T> Contains metrics for evaluating adversarial robustness of models.
RobustnessStats<T> Represents adversarial robustness diagnostics aggregated over a dataset.
RoundMetadata Contains detailed metrics for a single federated learning round.
SafetyFilterResult<T> Result of safety filtering on model output.
SafetyValidationResult<T> Result of safety validation for an input.
ScalarParameterSource<T> A single scalar exposed as a one-element parameter surface, optionally gated so it contributes nothing until the model is fitted.
SerializedObjectParameterSource<T> Exposes an explicitly classified fitted object graph through the canonical numeric parameter manifest without requiring its model to hand-write count, flatten and restore methods.
SplitChildLearningRateScales Per-attribute learning-rate multipliers applied to a Gaussian’s children immediately after it splits (#1835 excellence goal #4).
StageCallbacks<T, TInput, TOutput> Callbacks for training stage events.
TTestResult<T> Represents the results of a t-test, which is a statistical hypothesis test used to determine if there is a significant difference between the means of two groups.
TensorCollectionParameterSource<T> An index-stable tensor collection exposed as one parameter source.
TensorFieldParameterSource<T> A Tensor field exposed as a parameter surface, written through.
TensorListParameterSource<T> One or more tensor lists exposed as a single parameter surface, concatenated in the order the lists are given and, within each list, in index order.
TrainingRecipeReport What a model’s paper specifies for training, what is actually being used, and every difference between the two.
TrainingSpeedStats Contains statistics about training speed and progress.
TrainingStageResult<T, TInput, TOutput> Result of executing a training stage.
TrainingStage<T, TInput, TOutput> Represents a single stage in a training pipeline with comprehensive configuration options.
UncertaintyCalibrationData<TInput, TOutput> Provides optional calibration data for uncertainty quantification features.
UncertaintyPredictionResult<T, TOutput> Represents a prediction result augmented with uncertainty information.
UncertaintyStats<T> Represents uncertainty-quantification diagnostics aggregated over a dataset.
ValidationIssue Represents a specific validation issue.
VariableLengthParameterSource<T> Like DelegatingParameterSource, but sized BY the vector it is restored from.
VectorCollectionParameterSource<T> An index-stable vector collection exposed as one parameter source.
VectorFieldParameterSource<T> A whole Vector field exposed as a parameter surface.
VectorFieldWriteThroughSource<T> A Vector field exposed as a parameter surface, written through.
VectorModel<T> Represents a linear model that uses a vector of coefficients to make predictions.
VulnerabilityReport Detailed report of a specific vulnerability found during red teaming.

Base Classes (3)

Type Summary
ModelBase<T, TInput, TOutput> Abstract base class for standalone models that directly implement IFullModel.
ModelWrapperBase<T, TInput, TOutput> Abstract base class for model wrappers that delegate to an underlying IFullModel.
ParameterDistribution Base class for parameter distributions.

Interfaces (8)

Type Summary
IGeneratedParameterRegistrar<T> Implemented by generated partial classes so automated registration composes with hand-written exceptional registration instead of either surface suppressing the other.
IParameterChunkSource<T> Exposes the complete persistent state as role-aware chunks without conflating it with the trainable-only optimizer view.
IParameterLayoutSource Implemented by sources that can describe their local slots without allocating storage.
IParameterManifestProvider Implemented by models and layers that expose the generated parameter manifest.
IParameterMaterializationSource Implemented by lazy parameter sources that can make every shape-resolved slot writable.
IParameterSurfaceLifecycle Implemented by parameter sources whose structure or storage becomes ready in lifecycle stages.
IParameterTopologySource Persists the non-numeric structure that gives a variable parameter surface its shape.
IVariableLengthParameterSource<T> A component whose width is decided BY the vector it is restored from, not checked against it.

Enums (107)

Type Summary
ActiveLearningStrategyType Specifies the active learning strategy to use for sample selection.
AdaBoostAlgorithm AdaBoost algorithm variants.
AdvancedCompressionStrategy Advanced gradient compression strategies for federated learning communication efficiency.
AnomalyFeatures Flags for selecting which anomaly detection features to calculate.
AttestationPolicy Specifies the attestation policy for TEE remote attestation verification.
BackboneType Backbone network types for feature extraction.
BackdoorDetectionStrategy Specifies the backdoor detection strategy for federated learning.
BetaLinkFunction Link functions for Beta Regression mean model.
BlockSelectionMode Block selection mode for DFedBCA protocol.
ClassificationSplitCriterion Criterion used to measure the quality of a split in classification decision trees.
CloneConstructorArgumentKind The source of an explicitly bound constructor argument.
CloneCopyKind How a single configuration value is carried to a clone.
CloneMode How much of the original’s storage a copy is allowed to share.
ContinualLearningStrategyType Specifies the continual learning strategy to use for preventing catastrophic forgetting.
ContributionMethod Specifies the method used to evaluate client contributions in federated learning.
DARTDropoutMode Dropout modes for DART.
DARTDropoutType Types of dropout selection for DART.
DARTNormalization Normalization strategies for DART.
DARTNormalizationType Types of normalization after dropout in DART.
DecentralizedTopologyType Specifies the decentralized topology for peer-to-peer federated learning.
DetectionArchitecture Detection model architectures.
DifferencingFeatures Flags for selecting which differencing and stationarity features to compute.
DifferentialPrivacyMode Specifies where differential privacy noise is applied in the federated learning pipeline.
DistanceMetric Distance metrics for measuring similarity between samples.
EdgeHandling How to handle edge cases where the full window is not available.
FairnessConstraintType Specifies the type of fairness constraint to enforce during federated learning.
FederatedAdapterType Specifies the federated adapter type for parameter-efficient fine-tuning.
FederatedAggregationStrategy Specifies which federated aggregation strategy to use.
FederatedAsyncMode Specifies the asynchronous federated learning mode.
FederatedClientSelectionStrategy Specifies how clients are selected to participate in a federated training round.
FederatedCompressionStrategy Specifies the compression strategy for federated model updates.
FederatedContinualLearningStrategy Specifies the federated continual learning strategy for preventing catastrophic forgetting across rounds.
FederatedDistillationStrategy Specifies the federated knowledge distillation strategy.
FederatedDriftMethod Specifies the drift detection method used in federated learning.
FederatedHeterogeneityCorrection Specifies which heterogeneity correction algorithm to use.
FederatedLearningMode Specifies the federated learning paradigm to use.
FederatedMetaLearningStrategy Specifies the federated meta-learning strategy.
FederatedPersonalizationStrategy Specifies the personalization strategy for federated learning.
FederatedPrivacyAccountant Specifies which privacy accountant to use for reporting privacy spend in federated learning.
FederatedServerOptimizer Specifies which server-side federated optimizer (FedOpt family) to use.
FederatedStalenessWeighting Specifies how to down-weight stale updates in asynchronous federated learning.
FourierModeSelection How FEDformer chooses the frequency modes its Fourier blocks keep.
FrequencyAttentionType Specifies the type of frequency attention to use in FEDformer.
FuzzyMatchStrategy Specifies the similarity strategy used for fuzzy entity matching in PSI.
GAMLSSDistributionFamily Distribution families supported by GAMLSS.
GAMLSSModelType Types of sub-models available for modeling distribution parameters.
GLMMEstimationMethod Estimation methods for GLMM.
GLMMFamily Response distribution families for GLMM.
GLMMLinkFunction Link functions for GLMM.
GammaLinkFunction Link functions for Gamma regression.
GradientBoostingLoss Loss functions for Gradient Boosting classifier.
GradientClippingMethod Specifies the method used for gradient clipping.
GraphFLMode Specifies the federated graph learning task type.
GraphPartitionStrategy Specifies how a graph is partitioned across federated clients.
HomomorphicEncryptionMode Specifies how homomorphic encryption is applied during federated aggregation.
HomomorphicEncryptionScheme Specifies which homomorphic encryption scheme to use.
InverseGaussianLinkFunction Link functions for Inverse Gaussian regression.
KNNAlgorithm Algorithms for finding nearest neighbors.
LinearLoss Loss functions for linear classifiers.
LinearPenalty Penalty types for linear classifiers.
MissingFeatureStrategy Specifies how to handle missing feature blocks when not all parties have data for all entities.
MixedEffectsCovarianceStructure Covariance structures for random effects.
MixedEffectsEstimationMethod Estimation methods for mixed-effects models.
MixedEffectsOptimization Optimization methods for mixed-effects models.
MixedEffectsOptimizer Optimization algorithms for mixed-effects model fitting.
ModelSize Model size variants.
MpcProtocol Specifies the multi-party computation protocol to use.
MpcSecurityModel Specifies the adversary model assumed by the MPC protocol.
NGBoostDistributionType Types of distributions supported by NGBoost.
NHiTSInterpolationMode Interpolation method for upsampling forecasts in N-HiTS stacks.
NHiTSPoolingMode Pooling strategy for downsampling in N-HiTS stacks.
NeckType Neck architecture types for multi-scale feature fusion.
NmsType NMS (Non-Maximum Suppression) algorithm variants.
OrdinalLinkFunction Link functions for ordinal regression.
ParameterAvailability Declares when a slot is expected to become available.
ParameterOwnership Declares whether this manifest owns the storage it names.
ParameterPersistence Declares whether a numeric state slot belongs in a durable checkpoint.
ParameterReadiness States whether a parameter layout can be inspected or restored without allocation.
ParameterSlotRole Describes how a numeric slot participates in model state and optimization.
ParameterSurfaceIntent Identifies why a caller is preparing a parameter surface.
ParameterUpdatePolicy Declares which mechanism is allowed to change a numeric state slot.
PseudoNodeStrategy Specifies how to handle missing cross-client neighbor nodes in subgraph-level FL.
PsiProtocol Specifies the cryptographic protocol used for Private Set Intersection.
RecipeFidelity How faithfully a model’s training configuration reproduces the recipe its paper specifies.
RegularizationOrder Specifies which side of gradient clipping an optimizer adds its regularization on.
ReplayBufferStrategy Specifies the buffer management strategy for Experience Replay.
RollingRegressionFeatures Flags for selecting which rolling regression features to calculate.
RollingStatistics Flags for selecting which rolling statistics to calculate.
SeasonalityFeatures Flags for selecting which seasonality and calendar features to generate.
SecureAggregationMode Determines which secure aggregation protocol variant is used.
SplitPointStrategy Specifies how to choose the split point in a split neural network for vertical FL.
SuperLearnerMetaLearner Meta-learner types for Super Learner.
TechnicalIndicators Flags for selecting which technical indicators to calculate.
TeeProviderType Specifies the Trusted Execution Environment hardware provider.
TrialStatus Represents the status of a hyperparameter trial.
TweedieLinkFunction Link functions for Tweedie regression.
UnlearningMethod Specifies the federated unlearning method to use when a client requests data removal.
VerificationLevel Specifies the level of zero-knowledge verification applied to client updates.
VflAggregationMode Specifies how embeddings from multiple parties are combined in vertical federated learning.
VflUnlearningMethod Specifies the method used to remove an entity’s influence from a trained VFL model.
VolatilityMeasures Flags for selecting which volatility measures to calculate.
WeightingScheme Weighting schemes for neighbor voting.
WindowAutoDetectionMethod Methods for auto-detecting optimal window sizes.
ZeroInflatedCountLink Link functions for the count component of zero-inflated models.
ZeroInflatedDistributionFamily Base count distributions for zero-inflated models.
ZeroInflatedZeroLink Link functions for the zero-inflation component.
ZkProofSystem Specifies the zero-knowledge proof system to use for verifiable federated learning.

Structs (1)

Type Summary
ModelResult<T, TInput, TOutput> Represents the complete results of a model-building process, including the model solution, fitness metrics, fit detection results, evaluation data, and selected features.

Options & Configuration (493)

Type Summary
A2COptions<T> Configuration options for Advantage Actor-Critic (A2C) agents.
A3COptions<T> Configuration options for Asynchronous Advantage Actor-Critic (A3C) agents.
ABCNetOptions<T> Configuration for ABCNet.
ADMMOptimizerOptions<T, TInput, TOutput> Configuration options for the Alternating Direction Method of Multipliers (ADMM) optimization algorithm, which is particularly effective for problems with complex regularization requirements.
AIMOptions<T> Configuration options for AIM (Adaptive Iterative Mechanism), a marginal-based differentially private synthetic data generation method.
AMSGradOptimizerOptions<T, TInput, TOutput> Configuration options for the AMSGrad optimization algorithm, which is an improved variant of the Adam optimizer that addresses potential convergence issues by maintaining the maximum of past squared gradients.
ARIMAOptions<T> Configuration options for the ARIMA (AutoRegressive Integrated Moving Average) time series forecasting model.
ARIMAXModelOptions<T> Configuration options for the ARIMAX (AutoRegressive Integrated Moving Average with eXogenous variables) time series forecasting model.
ARMAOptions<T> Configuration options for the ARMA (AutoRegressive Moving Average) time series forecasting model.
ARModelOptions<T> Configuration options for the AR (AutoRegressive) time series forecasting model.
ASGDOptimizerOptions<T, TInput, TOutput> Configuration options for ASGD (Averaged Stochastic Gradient Descent), which runs ordinary decayed SGD while maintaining a running average of the iterates and uses that average as the answer.
ActiveLearningOptions Represents configuration options for active learning.
ActiveSetQuadraticProgramSolverOptions Configuration for the active-set quadratic-programming solver.
AdaBoostClassifierOptions<T> Configuration options for AdaBoost classifier.
AdaBoostR2RegressionOptions Configuration options for the AdaBoost R2 regression algorithm.
AdaDeltaOptimizerOptions<T, TInput, TOutput> Configuration options for the AdaDelta optimization algorithm, which is an extension of AdaGrad that adapts learning rates based on a moving window of gradient updates.
AdaMaxOptimizerOptions<T, TInput, TOutput> Configuration options for the AdaMax optimization algorithm, a variant of Adam that uses the infinity norm.
AdafactorOptimizerOptions<T, TInput, TOutput> Settings for AdafactorOptimizer.
AdagradOptimizerOptions<T, TInput, TOutput> Configuration options for the Adagrad optimization algorithm, which adapts the learning rate for each parameter based on historical gradient information.
Adam8BitOptimizerOptions<T, TInput, TOutput> Configuration options for the 8-bit Adam optimization algorithm, which reduces memory usage by quantizing optimizer states.
AdamOptimizerOptions<T, TInput, TOutput> Configuration options for the Adam optimization algorithm, which combines the benefits of AdaGrad and RMSProp.
AdamWOptimizerOptions<T, TInput, TOutput> Configuration options for the AdamW optimization algorithm with decoupled weight decay.
AdaptiveFitDetectorOptions Configuration options for the Adaptive Fit Detector, which automatically selects the most appropriate method to detect overfitting and underfitting in machine learning models.
AdaptiveRandomForestOptions<T> Configuration options for Adaptive Random Forest classifier.
AdvancedCompressionOptions Configuration for advanced gradient compression methods (PowerSGD, sketching, adaptive).
AdversarialAttackOptions<T> Configuration options for adversarial attack algorithms.
AdversarialDefenseOptions<T> Configuration options for adversarial defense mechanisms.
AdversarialRobustnessConfiguration<T> Non-generic version for backward compatibility and simpler use cases.
AdversarialRobustnessConfiguration<T, TInput, TOutput> Configuration for adversarial robustness and AI safety during model building and inference.
AdversarialRobustnessOptions<T> Unified configuration options for adversarial robustness and AI safety.
AiModelResultOptions<T, TInput, TOutput> Represents the configuration options for creating a AiModelResult.
AlgebraicRiccatiSolverOptions Configuration for the algebraic Riccati equation solvers.
AlignmentMethodOptions<T> Configuration options for AI alignment methods.
AlphaFactorOptions<T> Configuration options for the AlphaFactorModel.
AntColonyOptimizationOptions<T, TInput, TOutput> Configuration options for the Ant Colony Optimization algorithm, which is inspired by the foraging behavior of ants to find optimal paths through a search space.
AsyncFederatedLearningOptions Configuration options for asynchronous federated learning (FedAsync / FedBuff).
AudioHyperparameterOptions Shared configuration for audio and speech models: text-to-speech, speech recognition, speech enhancement and denoising, audio classification and audio generation.
AudioNeuralNetworkOptions Base configuration options for audio neural network models.
AugmentedLagrangianSolverOptions Configuration for the augmented Lagrangian solver.
AutoAttackOptions<T> Configuration for the AutoAttack ensemble (Croce and Hein 2020).
AutoCheckpointingOptions Configuration options for automatic checkpointing during training, supporting step-frequency, monotonic wall-clock interval, and metric-improvement triggers.
AutoDiffTabOptions<T> Configuration options for AutoDiff-Tab, an automated diffusion model for tabular data that searches over diffusion configurations and noise schedules to find optimal settings.
AutoIntOptions<T> Configuration options for AutoInt (Automatic Feature Interaction Learning).
AutocorrelationFitDetectorOptions Configuration options for detecting autocorrelation in time series data and regression residuals.
AutoformerOptions<T> Configuration options for the Autoformer model (Autoformer: Decomposition Transformers with Auto-Correlation for Long-Term Series Forecasting).
BFGSOptimizerOptions<T, TInput, TOutput> Configuration options for the BFGS (Broyden-Fletcher-Goldfarb-Shanno) optimization algorithm.
BackdoorDefenseOptions Configuration options for backdoor attack detection and mitigation in federated learning.
BayesianFitDetectorOptions Configuration options for the Bayesian model fit detector, which evaluates how well a model fits the data.
BayesianNetworkSynthOptions<T> Configuration options for Bayesian Network Synthesis, a statistical approach that learns a directed acyclic graph (DAG) structure and conditional probability tables to generate synthetic tabular data via ancestral sampling.
BayesianOptimizerOptions<T, TInput, TOutput> Configuration options for the Bayesian optimization algorithm.
BayesianRegressionOptions<T> Configuration options for Bayesian regression algorithms.
BayesianStructuralTimeSeriesOptions<T> Configuration options for Bayesian Structural Time Series models.
BetaRegressionOptions Configuration options for Beta Regression models.
BlackLittermanNeuralOptions<T> Configuration options for Neural Black-Litterman.
BloombergGPTOptions<T> Configuration options for BloombergGPT-style financial language model.
BootstrapFitDetectorOptions Configuration options for the Bootstrap Fit Detector, which evaluates model fit quality using bootstrap resampling.
BranchAndBoundSolverOptions Configuration for the branch-and-bound integer-programming solver.
CCDMOptions<T> Configuration options for CCDM, the Channel-aware Contrastive Conditional Diffusion model for multivariate probabilistic time series forecasting (Li, Chen and Xiong, 2024).
CLAPModelOptions Configuration options for CLAP (Contrastive Language-Audio Pretraining) models.
CMAESOptimizerOptions<T, TInput, TOutput> Configuration options for the CMA-ES (Covariance Matrix Adaptation Evolution Strategy) optimization algorithm.
CQLOptions<T> Configuration options for Conservative Q-Learning (CQL) agent.
CSDIOptions<T> Configuration options for CSDI (Conditional Score-based Diffusion model for Imputation).
CTABGANPlusOptions<T> Configuration options for CTAB-GAN+, an enhanced conditional tabular GAN with auxiliary classifier discriminator and mixed-type encoder for high-quality synthetic data generation.
CTGANOptions<T> Configuration options for CTGAN (Conditional Tabular GAN), a generative adversarial network specifically designed for generating realistic synthetic tabular data.
CalibratedClassifierOptions<T> Configuration options for calibrated classifiers.
CalibratedProbabilityFitDetectorOptions Configuration options for the Calibrated Probability Fit Detector, which evaluates how well a model’s predicted probabilities match actual outcomes.
CausalDiscoveryOptions Configuration options for causal structure discovery.
CausalGANOptions<T> Configuration options for Causal-GAN, a GAN that learns causal graph structure and generates data respecting causal relationships between features.
CertifiedDefenseOptions<T> Configuration options for certified defense mechanisms.
ChronosBoltOptions<T> Configuration options for Chronos-Bolt (Fast Non-Autoregressive Time Series Forecasting).
ChronosFinanceOptions<T> Configuration options for Finance.Forecasting.Foundation.Chronos, the tensor-API Chronos-T5 (Ansari et al.
ClassifierOptions<T> Configuration options for classification models, which are machine learning methods used to predict categorical outcomes (discrete classes) rather than continuous values.
ClientSelectionOptions Configuration options for client selection in federated learning.
CloneOptions Controls what a Clone carries across from the original.
CommitmentOptions Configuration options for cryptographic commitment schemes.
ConditionalInferenceTreeOptions Configuration options for Conditional Inference Trees, a statistically-driven approach to decision tree learning.
ConfusionMatrixFitDetectorOptions Configuration options for the Confusion Matrix Fit Detector, which evaluates how well a classification model performs.
ConjugateGradientOptimizerOptions<T, TInput, TOutput> Configuration options for the Conjugate Gradient optimization algorithm, which is used to train machine learning models.
ContinualLearningOptions Represents configuration options for continual learning.
ContributionEvaluationOptions Configuration options for client contribution evaluation in federated learning.
ConvTasNetOptions Configuration options for ConvTasNet audio source separation models.
CookDistanceFitDetectorOptions Configuration options for the Cook’s Distance fit detector, which helps identify influential data points and detect potential overfitting or underfitting in regression models.
CoordinateDescentOptimizerOptions<T, TInput, TOutput> Configuration options for the Coordinate Descent optimization algorithm, which optimizes a function by solving for one variable at a time while holding others constant.
CopulaGANOptions<T> Configuration options for CopulaGAN, a synthetic tabular data generator that combines Gaussian copula transformations with the CTGAN training pipeline.
CopulaSynthOptions<T> Configuration options for Copula-Based Synthesis, a statistical method that models the joint distribution of features by fitting marginal distributions individually and coupling them with a copula function.
CrossClientEdgeOptions Configuration for handling cross-client edges in federated graph learning.
CrossValidationFitDetectorOptions Configuration options for detecting overfitting, underfitting, and good fitting in machine learning models using cross-validation techniques.
CrossValidationOptions Represents the configuration options for cross-validation in machine learning models.
CrossformerOptions<T> Configuration options for the Crossformer model.
DARTClassifierOptions<T> Configuration options for DART (Dropouts meet Multiple Additive Regression Trees) classifier.
DARTOptions Configuration options for DART (Dropouts meet Multiple Additive Regression Trees).
DAdaptAdamOptimizerOptions<T, TInput, TOutput> Options for D-Adaptation Adam (Defazio and Mishchenko 2023), which estimates the step size from the iterates instead of taking it from a tuned learning rate.
DCCRNOptions Configuration options for DCCRN (Deep Complex Convolution Recurrent Network) audio enhancement models.
DCRNNOptions<T> Configuration options for DCRNN (Diffusion Convolutional Recurrent Neural Network).
DDPGOptions<T> Configuration options for DDPG agent.
DFPOptimizerOptions<T, TInput, TOutput> Configuration options for the Davidon-Fletcher-Powell (DFP) optimization algorithm, which is a quasi-Newton method used for finding local minima of functions.
DGCNNOptions Configuration options for DGCNN models.
DINOOptions<T> Hyperparameters of DINO (Zhang et al., “DINO: DETR with Improved DeNoising Anchor Boxes for End-to-End Object Detection”, ICLR 2023).
DLinearOptions<T> Options for DLinearModel — the decomposition-linear forecaster (Zeng et al., “Are Transformers Effective for Time Series Forecasting?”, AAAI 2023).
DPCTGANOptions<T> Configuration options for DP-CTGAN, a differentially private version of CTGAN that provides formal privacy guarantees while generating synthetic tabular data.
DQNOptions<T> Configuration options for Deep Q-Network (DQN) agents.
DecentralizedFederatedOptions Configuration options for decentralized (serverless) federated learning.
DecisionTransformerOptions<T> Configuration options for Decision Transformer agents.
DecisionTreeClassifierOptions<T> Configuration options for decision tree classifiers.
DecisionTreeOptions Configuration options for decision tree algorithms.
DeepAROptions<T> Configuration options for the DeepAR (Deep Autoregressive) model.
DeepFactorOptions<T> Configuration options for the DeepFactor (Deep Factor Model) for time series forecasting.
DeepFilterNetOptions Configuration options for DeepFilterNet audio enhancement models.
DeepHitOptions<T> Configuration options for DeepHit survival analysis model.
DeepPortfolioManagerOptions<T> Configuration options for the DeepPortfolioManager model.
DeepStateOptions<T> Configuration options for the DeepState (Deep State Space) model.
DeepSurvOptions<T> Configuration options for DeepSurv survival analysis model.
DetectionBackboneOptions Options shared by the detection backbones in src/ComputerVision/Detection/Backbones — the feature extractors (ResNet, EfficientNet, SwinTransformer, CSPDarknet) that a detector sits on top of.
DifferentialEvolutionOptions<T, TInput, TOutput> Configuration options for Differential Evolution optimization, a powerful variant of genetic algorithms that is particularly effective for continuous optimization problems.
DiffusionModelOptions<T> Configuration options for diffusion-based generative models.
DiffusionTSOptions<T> Configuration options for DiffusionTS (Interpretable Diffusion for Time Series).
DocumentNeuralNetworkOptions Shared configuration for document models: OCR and text recognition, layout analysis, document question answering, table and chart understanding, and document VQA.
DoubleDQNOptions<T> Configuration options for Double DQN agent.
DoubleQLearningOptions<T> Configuration options for Double Q-Learning agents.
DreamerOptions<T> Configuration options for Dreamer agents (Hafner et al.
DuelingDQNOptions<T> Configuration options for Dueling DQN agent.
DynaQOptions<T> Configuration options for Dyna-Q reinforcement learning agents.
DynaQPlusOptions<T> Configuration options for the Dyna-Q+ agent, which plans with a learned model and rewards revisiting long-untried actions.
DynamicRegressionWithARIMAErrorsOptions<T> Configuration options for Dynamic Regression with ARIMA Errors, a powerful time series forecasting method that combines regression with time series error correction.
ETSformerOptions<T> Configuration options for the ETSformer model.
EarlyStoppingConfig Early stopping configuration for a training stage.
ElasticNetRegressionOptions<T> Configuration options for Elastic Net Regression (combined L1 and L2 regularization).
EnsembleFitDetectorOptions Configuration options for the Ensemble Fit Detector, which combines multiple model fitness detectors to provide more robust and accurate recommendations for algorithm selection.
EpsilonGreedyBanditOptions<T> Configuration options for the epsilon-greedy multi-armed bandit.
ExpectedSARSAOptions<T> Configuration options for Expected SARSA agents.
ExplainableBoostingClassifierOptions<T> Configuration options for Explainable Boosting Machine (EBM) classifier.
ExplainableBoostingMachineOptions Configuration options for Explainable Boosting Machine (EBM) models.
ExponentialSmoothingOptions<T> Configuration options for Exponential Smoothing, a time series forecasting method that gives exponentially decreasing weights to older observations.
ExtraTreesClassifierOptions<T> Configuration options for Extra Trees (Extremely Randomized Trees) classifier.
ExtremelyRandomizedTreesRegressionOptions Configuration options for Extremely Randomized Trees regression, an ensemble learning method that builds multiple decision trees with additional randomization for improved prediction accuracy.
FEDformerOptions<T> Configuration options for the FEDformer (Frequency Enhanced Decomposed Transformer) model.
FTRLOptimizerOptions<T, TInput, TOutput> Configuration options for the Follow-The-Regularized-Leader (FTRL) optimizer, an advanced gradient-based optimization algorithm particularly effective for sparse datasets and online learning.
FTTransformerOptions<T> Configuration options for FT-Transformer, a Feature Tokenizer + Transformer for tabular data.
FactorVAEOptions<T> Configuration options for the FactorVAE model.
FeatureImportanceFitDetectorOptions Configuration options for the Feature Importance Fit Detector, which analyzes how different input features contribute to a model’s predictions and evaluates potential issues with model fit.
FederatedAdapterOptions Configuration options for federated parameter-efficient fine-tuning (PEFT).
FederatedCompressionOptions Configuration options for federated update compression (quantization, sparsification, and error feedback).
FederatedContinualLearningOptions Configuration options for federated continual learning (preventing catastrophic forgetting in FL).
FederatedDistillationOptions Configuration options for federated knowledge distillation, enabling model-heterogeneous FL.
FederatedDriftOptions Configuration options for federated concept drift detection and adaptation.
FederatedFairnessOptions Configuration options for fairness constraints in federated learning.
FederatedGraphOptions Configuration options for federated graph learning.
FederatedHeterogeneityCorrectionOptions Configuration options for federated heterogeneity correction algorithms.
FederatedLearningOptions Configuration options for federated learning training.
FederatedMetaLearningOptions Configuration options for federated meta-learning.
FederatedPersonalizationOptions Configuration options for personalized federated learning (PFL).
FederatedServerOptimizerOptions Configuration options for server-side federated optimizers (FedOpt family).
FederatedUnlearningOptions Configuration options for federated unlearning (right to be forgotten).
FinBERTOptions<T> Configuration options for FinBERT (Financial BERT) model.
FinBERTToneOptions<T> Configuration options for FinBERT-Tone model for fine-grained financial sentiment/tone analysis.
FinDiffOptions<T> Configuration options for FinDiff, a diffusion model specialized for generating realistic financial tabular data with temporal correlation preservation.
FinGPTOptions<T> Configuration options for FinGPT (Financial GPT) model.
FinMAOptions<T> Configuration options for FinMA (Financial Multi-Agent) model.
FinRLAgentOptions<T> Configuration options for the FinRL unified trading agent.
FinancialA2CAgentOptions<T> Configuration options for the Financial A2C (Advantage Actor-Critic) trading agent.
FinancialAutoMLOptions<T> Configuration options for financial AutoML runs.
FinancialBERTOptions<T> Configuration options for FinancialBERT model - a domain-adapted BERT for comprehensive financial analysis.
FinancialDQNAgentOptions<T> Configuration options for the Financial DQN (Deep Q-Network) trading agent.
FinancialNLPOptions Base configuration options for financial NLP models.
FinancialNeuralNetworkOptions Base configuration options for financial neural network models.
FinancialPPOAgentOptions<T> Configuration options for the Financial PPO (Proximal Policy Optimization) trading agent.
FinancialSACAgentOptions<T> Configuration options for the Financial SAC (Soft Actor-Critic) trading agent.
FineTuningConfiguration<T, TInput, TOutput> Configuration for fine-tuning during model building.
FineTuningOptions<T> Configuration options for fine-tuning methods.
FitnessCalculatorOptions Configuration options for the fitness calculator, which determines how model performance is evaluated.
FlowStateOptions<T> Configuration options for FlowState (IBM’s SSM-based Time Series Foundation Model).
ForecastingModelOptions Base configuration options for financial forecasting models.
FuzzyMatchOptions Configuration options for fuzzy entity matching in Private Set Intersection.
GAMLSSOptions Configuration options for GAMLSS (Generalized Additive Models for Location, Scale, and Shape).
GANDALFOptions<T> Configuration options for GANDALF (Gated Additive Neural Decision Forest).
GARCHModelOptions<T> Configuration options for the Generalized Autoregressive Conditional Heteroskedasticity (GARCH) model, which is used for analyzing and forecasting volatility in time series data.
GLMMOptions<T> Configuration options for Generalized Linear Mixed-Effects Models (GLMM).
GOGGLEOptions<T> Configuration options for GOGGLE (Generative mOdelling for tabular data by learninG reLational structurE), a graph-based VAE that learns feature dependency structure for generating realistic tabular data.
GPT4TSOptions<T> Configuration options for GPT4TS (One Fits All: Power General Time Series Analysis by Pretrained LM).
GammaRegressionOptions<T> Configuration options for Gamma Regression, a generalized linear model for positive continuous data.
GaussianCompressionOptions Per-technique configuration for the GaussianSplatting post-training compression pass (#1835 excellence goal #3).
GaussianProcessFitDetectorOptions Configuration options for the Gaussian Process Fit Detector, which analyzes model fit quality using Gaussian Process regression to detect overfitting, underfitting, and uncertainty issues.
GaussianProcessRegressionOptions Configuration options for Gaussian Process Regression, a flexible non-parametric approach to regression that provides uncertainty estimates along with predictions.
GaussianSplattingOptions Configuration options for Gaussian Splatting models (Kerbl et al.
GeneralizedAdditiveModelOptions<T> Configuration options for Generalized Additive Models (GAMs), which are flexible regression models that combine multiple simple functions to model complex relationships.
GeneticAlgorithmOptimizerOptions<T, TInput, TOutput> Configuration options for genetic algorithm optimization, which uses principles inspired by natural selection to find optimal solutions to complex problems.
GlobalEarlyStoppingConfig Global early stopping configuration that spans multiple training stages.
GradientBanditOptions<T> Configuration options for the gradient bandit, which learns action preferences with a softmax policy.
GradientBasedOptimizerOptions<T, TInput, TOutput> Configuration options for gradient-based optimization algorithms.
GradientBoostingClassifierOptions<T> Configuration options for Gradient Boosting classifier.
GradientBoostingFitDetectorOptions Configuration options for the Gradient Boosting Fit Detector, which analyzes model fit quality to detect overfitting in gradient boosting models.
GradientBoostingRegressionOptions Configuration options for Gradient Boosting Regression, an ensemble learning technique that combines multiple decision trees to create a powerful regression model.
GradientDescentOptimizerOptions<T, TInput, TOutput> Configuration options for the Gradient Descent optimizer, which is a fundamental algorithm for finding the minimum of a function by iteratively moving in the direction of steepest descent.
GraphAttentionPortfolioOptions<T> Configuration for GraphAttentionPortfolio.
GraphEncoderOptions Shared hyperparameters for the graph encoder networks — GraphSAGE, the Graph Isomorphism Network and the Graph Attention Network.
GraphModelOptions Shared hyperparameters for the graph task models — the node, link and graph-level heads that sit on a message-passing encoder.
GraphWaveNetOptions<T> Configuration options for GraphWaveNet (Graph WaveNet for Deep Spatial-Temporal Modeling).
HeteroscedasticityFitDetectorOptions Configuration options for the Heteroscedasticity Fit Detector, which analyzes whether a model’s prediction errors have constant variance across all prediction values.
HierarchicalRiskParityOptions<T> Configuration options for Hierarchical Risk Parity.
HippoOptions<T> Configuration options for HiPPO (High-order Polynomial Projection Operators) model.
HistGradientBoostingOptions Configuration options for Histogram-based Gradient Boosting, a fast ensemble learning technique that uses binned features for efficient tree building.
HoeffdingTreeOptions<T> Configuration options for Hoeffding Tree classifier.
HoldoutValidationFitDetectorOptions Configuration options for the Holdout Validation Fit Detector, which analyzes model performance on separate training and validation datasets to identify overfitting, underfitting, and other model quality issues.
HomomorphicEncryptionOptions Configuration options for homomorphic encryption (HE) in federated learning.
HybridFitDetectorOptions Configuration options for the Hybrid Fit Detector, which combines multiple model evaluation techniques to provide a comprehensive assessment of model quality.
IQLOptions<T> Configuration options for Implicit Q-Learning (IQL) agent.
ITransformerOptions<T> Configuration options for the iTransformer (Inverted Transformer) model.
InformationCriteriaFitDetectorOptions Configuration options for the Information Criteria Fit Detector, which uses statistical information criteria like AIC and BIC to evaluate model quality and complexity trade-offs.
InformerOptions<T> Configuration options for the Informer model (Informer: Beyond Efficient Transformer for Long Sequence Time-Series Forecasting).
InstantNGPOptions<T> Configuration options for Instant-NGP models.
InteriorPointSolverOptions Configuration for the primal-dual interior-point solver.
InterpretabilityOptions Configuration options for model interpretability and explainability features.
InterventionAnalysisOptions<T, TInput, TOutput> Configuration options for Intervention Analysis, which is a time series modeling technique used to assess the impact of specific events or interventions on a time series.
InverseGaussianRegressionOptions<T> Configuration options for Inverse Gaussian Regression, a generalized linear model for positive continuous data with variance proportional to the cube of the mean.
InvestLMOptions<T> Configuration options for InvestLM (Investment Language Model).
JackknifeFitDetectorOptions Configuration options for the Jackknife Fit Detector, which uses the jackknife resampling technique to evaluate model stability and detect overfitting or underfitting.
KFoldCrossValidationFitDetectorOptions Configuration options for the K-Fold Cross Validation Fit Detector, which evaluates model quality by analyzing performance across multiple data partitions.
KNearestNeighborsOptions Configuration options for the K-Nearest Neighbors algorithm, which makes predictions based on the values of the K closest data points in the training set.
KNeighborsOptions<T> Configuration options for K-Nearest Neighbors classifiers.
KairosOptions<T> Configuration options for Kairos (Adaptive and Generalizable Time Series Foundation Model).
KernelRidgeRegressionOptions Configuration options for Kernel Ridge Regression, which combines ridge regression with the kernel trick to model non-linear relationships in data.
KnowledgeDistillationOptions<T, TInput, TOutput> Configuration options for knowledge distillation training.
KronosOptions<T> Configuration options for Kronos (Foundation Model for the Language of Financial Markets).
LAMBOptimizerOptions<T, TInput, TOutput> Configuration options for the LAMB (Layer-wise Adaptive Moments for Batch training) optimization algorithm.
LARSOptimizerOptions<T, TInput, TOutput> Configuration options for the LARS (Layer-wise Adaptive Rate Scaling) optimization algorithm.
LBFGSOptimizerOptions<T, TInput, TOutput> Configuration options for the Limited-memory Broyden-Fletcher-Goldfarb-Shanno (L-BFGS) optimizer, which is an efficient optimization algorithm for training machine learning models.
LLMTimeOptions<T> Configuration options for LLM-Time (Zero-Shot Time Series Forecasting via LLM Tokenization).
LSPIOptions<T> Configuration options for LSPI (Least-Squares Policy Iteration) agents.
LSTDOptions<T> Configuration options for LSTD (Least-Squares Temporal Difference) agents.
LSTNetOptions<T> Configuration options for the LSTNet (Long Short-Term Time-series Network) model.
LagLlamaOptions<T> Configuration options for Lag-Llama (Large Language Model for time series forecasting).
LassoRegressionOptions<T> Configuration options for Lasso Regression (L1 regularized linear regression).
LearningCurveFitDetectorOptions Configuration options for the Learning Curve Fit Detector, which analyzes training progress to determine when a model has converged or is unlikely to improve further.
LevenbergMarquardtOptimizerOptions<T, TInput, TOutput> Configuration options for the Levenberg-Marquardt optimization algorithm, which is used for non-linear least squares optimization in machine learning and AI models.
LinearClassifierOptions<T> Configuration options for linear classifiers.
LinearMatrixInequalityOptions Configuration for the linear matrix inequality solver.
LinearMixedModelOptions<T> Configuration options for mixed-effects (hierarchical/multilevel) models.
LinearQLearningOptions<T> Configuration options for Linear Q-Learning agents.
LinearSARSAOptions<T> Configuration options for Linear SARSA agents.
LionOptimizerOptions<T, TInput, TOutput> Configuration options for the Lion (Evolved Sign Momentum) optimization algorithm.
LoRAConfiguration LoRA configuration for parameter-efficient fine-tuning.
LocallyWeightedRegressionOptions Configuration options for Locally Weighted Regression, a non-parametric method that creates a model by fitting simple models to localized subsets of data.
LogisticRegressionOptions<T> Configuration options for Logistic Regression, a statistical method used for binary classification problems in machine learning.
M5ModelTreeOptions Configuration options for the M5 Model Tree algorithm, which combines decision trees with linear regression models at the leaf nodes.
MADDPGOptions<T> Configuration options for Multi-Agent DDPG (MADDPG) agents.
MAModelOptions<T> Configuration options for Moving Average (MA) models, which are used to analyze time series data by modeling the error terms as a linear combination of previous error terms.
MGTSDOptions<T> Configuration options for MG-TSD (Multi-Granularity Time Series Diffusion).
MLkNNOptions<T> Configuration options for ML-kNN (Multi-Label k-Nearest Neighbors) classifier.
MOIRAIOptions<T> Configuration options for MOIRAI (Salesforce’s Universal Time Series Foundation Model).
MOMENTOptions<T> Configuration options for MOMENT (Multi-task Optimization through Masked Encoding for Time series) foundation model.
MQCNNOptions<T> Configuration options for the MQCNN (Multi-Quantile Convolutional Neural Network) model.
MTGNNOptions<T> Configuration options for MTGNN (Multivariate Time-series Graph Neural Network).
Mamba2Options<T> Configuration options for Mamba-2 (Structured State Space Duality) forecasting.
MambaOptions<T> Configuration options for Mamba (Selective State Space Model).
MambularOptions<T> Configuration options for Mambular (State Space Models for Tabular Data).
MarketMakingOptions<T> Configuration options for the MarketMakingAgent.
MaximalInformationOptions The MINE grid parameters of the maximal information coefficient (Reshef et al.
MedGANOptions<T> Configuration options for medGAN, the medical Generative Adversarial Network of Choi et al.
MeshCNNOptions Configuration options for the MeshCNN neural network.
MiniBatchGradientDescentOptions<T, TInput, TOutput> Configuration options for Mini-Batch Gradient Descent, an optimization algorithm that updates model parameters using the average gradient computed from small random subsets of training data.
MiniRocketOptions<T> Configuration options for the MiniRocket time series classifier.
MisGANOptions<T> Configuration options for MisGAN, a GAN for learning from incomplete data with dual generator/discriminator pairs for data and missingness patterns.
MissingFeatureOptions Configuration for handling missing features in vertical federated learning.
MixedEffectsModelOptions Configuration options for Mixed-Effects (Hierarchical/Multilevel) Models.
MixtureOfExpertsOptions<T> Configuration options for the Mixture-of-Experts (MoE) neural network model.
ModelHyperparameterOptions Base class for the per-model-family hyperparameter options: the sizes and shapes that come from the paper a model was published in.
ModelOptions Base class of every model’s options: the settings shared by all models, such as the random seed.
ModelPredictiveControllerOptions<T> Configuration for the linear model predictive controller.
ModelStatsOptions Configuration options for model statistics and diagnostics calculations, which help evaluate the quality, reliability, and performance of machine learning models.
ModifiedPolicyIterationOptions<T> Configuration options for Modified Policy Iteration agents.
MomentumOptimizerOptions<T, TInput, TOutput> Configuration options for the Momentum Optimizer, which enhances gradient descent by adding a fraction of the previous update direction to the current update.
MonteCarloExploringStartsOptions<T> Configuration options for Monte Carlo Exploring Starts agents.
MonteCarloOptions<T> Configuration options for Monte Carlo agents.
MonteCarloValidationOptions Represents the options for Monte Carlo cross-validation.
MpcOptions Configuration options for multi-party computation protocols in federated learning.
MuZeroOptions<T> Configuration options for MuZero agents.
MultilayerPerceptronOptions<T, TInput, TOutput> Configuration options for Multilayer Perceptron (MLP), a type of feedforward artificial neural network that consists of multiple layers of neurons.
MultinomialLogisticRegressionOptions<T> Configuration options for Multinomial Logistic Regression, a classification method that generalizes logistic regression to multiclass problems with more than two possible discrete outcomes.
NASOptions<T> Configuration options for Neural Architecture Search (NAS).
NBEATSModelOptions<T> Configuration options for the N-BEATS (Neural Basis Expansion Analysis for Time Series) model.
NGBoostClassifierOptions<T> Configuration options for NGBoost (Natural Gradient Boosting) classifier models.
NGBoostRegressionOptions<T> Configuration options for NGBoost (Natural Gradient Boosting) regression models.
NHiTSOptions<T> Configuration options for the N-HiTS (Neural Hierarchical Interpolation for Time Series) model.
NLinearOptions<T> Options for NLinearModel — the normalization-linear forecaster (Zeng et al., AAAI 2023).
NODEOptions<T> Configuration options for NODE (Neural Oblivious Decision Ensembles).
NStepQLearningOptions<T> Configuration options for n-step Q-learning.
NStepSARSAOptions<T> Configuration options for N-step SARSA agents.
NadamOptimizerOptions<T, TInput, TOutput> Configuration options for the Nadam optimizer, which combines Nesterov momentum with Adam’s adaptive learning rates for efficient training of neural networks and other gradient-based models.
NaiveBayesOptions<T> Configuration options for Naive Bayes classifiers.
NeRFOptions Configuration options for NeRF models.
NegativeBinomialRegressionOptions<T> Configuration options for Negative Binomial Regression, a statistical model used for count data that exhibits overdispersion (variance exceeding the mean).
NelderMeadOptimizerOptions<T, TInput, TOutput> Configuration options for the Nelder-Mead optimization algorithm, a derivative-free method for finding the minimum of an objective function in a multidimensional space.
NesterovAcceleratedGradientOptimizerOptions<T, TInput, TOutput> Configuration options for the Nesterov Accelerated Gradient optimization algorithm, a momentum-based technique that improves convergence speed in gradient descent optimization.
NeuralCVaROptions<T> Configuration options for the NeuralCVaR model.
NeuralGARCHOptions<T> Configuration options for the Neural GARCH volatility model.
NeuralNetworkARIMAOptions<T> Configuration options for Neural Network ARIMA (AutoRegressive Integrated Moving Average) models, which combine traditional statistical time series methods with neural networks for improved forecasting.
NeuralNetworkFitDetectorOptions Configuration options for the Neural Network Fit Detector, which evaluates the quality of a neural network’s fit to data by analyzing performance metrics and detecting issues like underfitting and overfitting.
NeuralNetworkOptions Base configuration options for all neural network models.
NeuralNetworkRegressionOptions<T, TInput, TOutput> Configuration options for neural network regression models, providing fine-grained control over network architecture, training parameters, activation functions, and optimization strategies.
NeuralNoiseReducerOptions Configuration options for neural noise reduction models.
NeuralStressTestOptions<T> Configuration options for the NeuralStressTest model.
NeuralVaROptions<T> Configuration options for the NeuralVaR model.
NewtonMethodOptimizerOptions<T, TInput, TOutput> Configuration options for Newton’s Method optimizer, an advanced second-order optimization technique that uses both gradient and Hessian information to accelerate convergence in optimization problems.
NonLinearRegressionOptions Configuration options for nonlinear regression models, which capture complex, nonlinear relationships between input features and output variables using kernel functions and iterative optimization.
NonStationaryTransformerOptions<T> Configuration options for the Non-stationary Transformer model.
NonlinearModelPredictiveControllerOptions<T> Configuration for the nonlinear model predictive controller.
OCTGANOptions<T> Configuration options for OCT-GAN (Kim et al.
ObjectDetectionOptions<T> Configuration options for object detection models.
OffPolicyMonteCarloOptions<T> Configuration options for Off-Policy Monte Carlo Control agents with importance sampling.
OhlcColumnConfig Configuration for OHLC (Open, High, Low, Close) column indices.
OnPolicyMonteCarloOptions<T> Configuration options for On-Policy Monte Carlo Control agents.
OnlineClassifierOptions<T> Base configuration options for online (incremental) classifiers.
OnlineNaiveBayesOptions<T> Configuration options for Online Naive Bayes classifier.
OptimizationAlgorithmOptions<T, TInput, TOutput> Configuration options for optimization algorithms used in machine learning models.
OrdinalRegressionOptions<T> Configuration options for Ordinal Regression (Proportional Odds Model), a classification method for predicting ordered categorical outcomes.
OrthogonalRegressionOptions<T> Configuration options for Orthogonal Regression (also known as Total Least Squares), which minimizes the perpendicular distances from data points to the fitted model, accounting for errors in both dependent and independent variables.
PANNsModelOptions Configuration options for PANNs (Pretrained Audio Neural Networks) models (Kong et al.
PATEGANOptions<T> Configuration options for PATE-GAN (Jordon, Yoon and van der Schaar, ICLR 2019), a GAN whose discriminator is a student trained on generated samples labelled by a noisy vote of teachers, each of which sees only one disjoint partition of the…
PPOOptions<T> Configuration options for Proximal Policy Optimization (PPO) agents.
PanopticSegmentationOptions Shared hyperparameters for the panoptic segmentation family — the query-based mask transformers that assign every pixel both a class and an instance.
PartialDependencePlotFitDetectorOptions Configuration options for the Partial Dependence Plot Fit Detector, which uses partial dependence plots to evaluate model fit quality and detect overfitting or underfitting in machine learning models.
PartialLeastSquaresRegressionOptions<T> Configuration options for Partial Least Squares Regression (PLS), a technique that combines features of principal component analysis and multiple regression to handle multicollinearity and high-dimensional data.
ParticleSwarmOptimizationOptions<T, TInput, TOutput> Configuration options for Particle Swarm Optimization (PSO), a population-based stochastic optimization technique inspired by social behavior of bird flocking or fish schooling.
PatchTSTOptions<T> Configuration options for the PatchTST (Patch Time Series Transformer) model.
PermutationTestFitDetectorOptions Configuration options for the permutation test fit detector, which helps identify overfitting, underfitting, and high variance in machine learning models.
PhysicsInformedOptions Base configuration options for physics-informed neural network models.
PointNetOptions Configuration options for PointNet models.
PointNetPlusPlusOptions Configuration options for PointNet++ models.
PoissonRegressionOptions<T> Configuration options for Poisson Regression, a specialized form of regression analysis used for modeling count data and contingency tables where the dependent variable consists of non-negative integers.
PolicyIterationOptions<T> Configuration options for Policy Iteration agents.
PolynomialRegressionOptions<T> Configuration options for Polynomial Regression, an extension of linear regression that models the relationship between variables using polynomial functions to capture non-linear relationships in data.
PortfolioOptimizerOptions<T> Base options for portfolio optimizers.
PortfolioOptions<T> Configuration options for portfolio optimization models.
PowellOptimizerOptions<T, TInput, TOutput> Configuration options for Powell’s method, a derivative-free optimization algorithm used for finding the minimum of a function without requiring gradient information.
PrecisionRecallCurveFitDetectorOptions Configuration options for the Precision-Recall Curve Fit Detector, which evaluates model quality using precision-recall metrics particularly valuable for imbalanced classification problems.
PredictionStatsOptions Configuration options for prediction statistics generation, which provides statistical analysis and reporting for model predictions including confidence intervals and learning curve analysis.
PrincipalComponentRegressionOptions<T> Configuration options for Principal Component Regression (PCR), which combines principal component analysis with linear regression to address multicollinearity and dimensionality issues in regression problems.
PrioritizedSweepingOptions<T> Configuration options for prioritized sweeping, which plans first where the value estimates would change most.
ProbabilityCalibratorOptions Configuration options for probability calibrator.
ProphetOptions<T, TInput, TOutput> Configuration options for Prophet, a procedure for forecasting time series data based on an additive model where non-linear trends are fit with yearly, weekly, and daily seasonality, plus holiday effects.
ProximalGradientDescentOptimizerOptions<T, TInput, TOutput> Configuration options for the Proximal Gradient Descent optimizer, an advanced optimization algorithm that combines traditional gradient descent with proximal operators to handle regularization effectively.
PsiOptions Configuration options for Private Set Intersection (PSI) protocols.
QLambdaOptions<T> Configuration options for Q(lambda), Q-learning with eligibility traces.
QMIXOptions<T> Configuration options for QMIX agents (Rashid et al.
QuantileRegressionForestsOptions Configuration options for Quantile Regression Forests, an extension of Random Forests that enables prediction of conditional quantiles rather than just the conditional mean.
QuantileRegressionOptions<T> Configuration options for Quantile Regression, a technique that enables prediction of specific quantiles of the conditional distribution rather than just the conditional mean.
RAdamOptimizerOptions<T, TInput, TOutput> Configuration options for the RAdam (Rectified Adam) optimization algorithm, which adds a variance rectification term to Adam so that the adaptive learning rate is only trusted once enough gradient history exists to estimate it reliably.
RAkELOptions<T> Configuration options for RAkEL (Random k-Labelsets) classifier.
REINFORCEOptions<T> Configuration options for REINFORCE agents.
REaLTabFormerOptions<T> Configuration options for REaLTabFormer, a GPT-2 style autoregressive transformer for generating realistic tabular data by treating columns as a sequence of tokens.
ROCCurveFitDetectorOptions Configuration options for the ROC Curve Fit Detector, which evaluates classification model quality using Receiver Operating Characteristic (ROC) curve analysis.
RTDETROptions<T> Hyperparameters of RT-DETR (Zhao et al., “DETRs Beat YOLOs on Real-time Object Detection”, CVPR 2024).
RWKV7LanguageModelOptions<T> Configuration options for the RWKV-7 “Goose” language model.
RWKVForecastingOptions<T> Configuration options for RWKV-based time series forecasting.
RadialBasisFunctionOptions Configuration options for Radial Basis Function (RBF) models, a type of artificial neural network that uses radial basis functions as activation functions for approximating complex non-linear relationships.
RainbowDQNOptions<T> Configuration options for Rainbow DQN agent.
RandomForestClassifierOptions<T> Configuration options for Random Forest classifiers.
RandomForestRegressionOptions Configuration options for Random Forest Regression, an ensemble learning method that combines multiple decision trees to improve prediction accuracy and control overfitting.
RealizedVolatilityTransformerOptions<T> Configuration options for the Realized Volatility Transformer model.
ReferringSegmentationOptions Options shared by the referring segmentation models (LISA, VideoLISA, OMG-LLaVA, PixelLM), which turn a referring expression into one text embedding and decode a binary mask from it.
RegressionOptions<T> Configuration options for regression models, which are statistical methods used to estimate relationships between variables and make predictions.
RegularizationOptions Configuration options for regularization techniques used to prevent overfitting in machine learning models.
RelationalGCNOptions<T> Configuration options for RelationalGCN (Relational Graph Convolutional Network).
ReservoirOptions Shared hyperparameters for the reservoir computing models — the Echo State Network and the Liquid State Machine.
ResidualAnalysisFitDetectorOptions Configuration options for the Residual Analysis Fit Detector, which evaluates model fit quality by analyzing prediction residuals against various statistical thresholds.
ResidualBootstrapFitDetectorOptions Configuration options for the Residual Bootstrap Fit Detector, which uses bootstrap resampling of residuals to assess model fit quality and detect overfitting or underfitting.
RidgeRegressionOptions<T> Configuration options for Ridge Regression (L2 regularized linear regression).
RiskModelOptions<T> Base options for risk models.
RobustAggregationOptions Configuration options for robust aggregation strategies in federated learning.
RobustRegressionOptions<T> Configuration options for robust regression models, which are designed to be less sensitive to outliers and violations of standard regression assumptions.
RootMeanSquarePropagationOptimizerOptions<T, TInput, TOutput> Configuration options for the Root Mean Square Propagation (RMSProp) optimizer, an adaptive learning rate optimization algorithm commonly used in training neural networks.
RpropOptimizerOptions<T, TInput, TOutput> Configuration options for Rprop (resilient backpropagation), which adapts a separate step size for every weight from the SIGN of its gradient and discards the magnitude entirely.
S4Options<T> Configuration options for S4 (Structured State Space Sequence Model).
SACOptions<T> Configuration options for Soft Actor-Critic (SAC) agents.
SAINTOptions<T> Configuration options for SAINT (Self-Attention and Intersample Attention Transformer).
SARIMAOptions<T> Configuration options for Seasonal Autoregressive Integrated Moving Average (SARIMA) models, which extend ARIMA models to incorporate seasonal components in time series data.
SARSALambdaOptions<T> Configuration options for SARSA(lambda), on-policy TD control with eligibility traces.
SARSAOptions<T> Configuration options for SARSA agents.
SECBERTOptions<T> Configuration options for SEC-BERT model specialized for SEC filings analysis.
SMOTENCOptions<T> Configuration options for SMOTE-NC (Synthetic Minority Over-sampling Technique for Nominal and Continuous features), a k-NN based oversampling method that generates synthetic minority samples by interpolating between existing ones.
STGNNOptions<T> Configuration options for STGNN (Spatio-Temporal Graph Neural Network).
STLDecompositionOptions<T> Configuration options for Seasonal-Trend-Loess (STL) decomposition, a versatile method for decomposing time series into seasonal, trend, and residual components.
SVMOptions<T> Configuration options for Support Vector Machine classifiers.
SafetyFilterConfiguration<T> Configuration for safety filtering during inference.
SafetyFilterOptions<T> Configuration options for safety filtering mechanisms.
ScaledAdamOptimizerOptions<T, TInput, TOutput> Configuration for ScaledAdamOptimizer.
ScheduleFreeAdamWOptimizerOptions<T, TInput, TOutput> Settings for ScheduleFreeAdamWOptimizer.
ScoreGradOptions<T> Configuration options for ScoreGrad (Score-based Gradient Models for Time Series).
SecureAggregationOptions Configuration options for secure aggregation in federated learning (Bonawitz et al.
SegmentationModelOptions The hyperparameters every segmentation model in the library has: how many classes it predicts, and how much dropout its encoder uses.
SequentialMinimalOptimizationOptions Configuration for the Sequential Minimal Optimization solver used to train support-vector machines.
ShapleyValueFitDetectorOptions Configuration options for the Shapley Value Fit Detector, which evaluates model fit quality by analyzing feature importance using Shapley values.
SignatureInformedTransformerOptions<T> Configuration for SignatureInformedTransformer.
SileroVadOptions Configuration options for Silero VAD (Voice Activity Detection) models.
SimMTMOptions<T> Configuration options for SimMTM (Simple Pre-Training Framework for Masked Time-Series Modeling).
SimplexSolverOptions Configuration for the simplex linear-programming solver.
SimulatedAnnealingOptions<T, TInput, TOutput> Configuration options for the Simulated Annealing optimization algorithm, a probabilistic technique for approximating the global optimum of a given function.
SpeakerRecognitionOptions Shared hyperparameters for the speaker recognition models — the embedding extractor and the verifier built on top of it.
SpectralAnalysisOptions<T> Configuration options for spectral analysis of time series data, which transforms time-domain signals into the frequency domain to identify periodic components and patterns.
SpiralNetOptions Configuration options for SpiralNet++ mesh neural network.
SplineRegressionOptions Configuration options for spline regression models, which fit piecewise polynomial functions to data for flexible nonlinear modeling.
SplitModelOptions Configuration for split neural network architecture in vertical federated learning.
StateSpaceModelOptions<T> Configuration options for State Space Models, which represent time series data through hidden states and observable outputs for forecasting and analysis.
StepwiseRegressionOptions<T> Configuration options for Stepwise Regression, an automated feature selection approach that iteratively adds or removes predictors based on their statistical significance.
StochasticGradientDescentOptimizerOptions<T, TInput, TOutput> Configuration options for Stochastic Gradient Descent (SGD) optimization, a widely used algorithm for training machine learning models with large datasets.
StockformerOptions<T> Configuration for Stockformer.
StratifiedKFoldCrossValidationFitDetectorOptions Configuration options for detecting overfitting, underfitting, and model stability using stratified k-fold cross-validation.
StreamingTrainingOptions<T, TInput, TOutput> Controls step-level behavior of the streaming training path (ConfigureDataLoader with an IStreamingDataLoader): a reproducible data order, a global step budget, periodic held-out validation, and resuming from the latest checkpoint.
SubgraphSamplingOptions Configuration for subgraph neighborhood sampling during federated GNN training.
SundialOptions<T> Configuration options for Sundial (A Family of Highly Capable Time Series Foundation Models).
SuperLearnerOptions Configuration options for Super Learner (Stacking) ensemble.
SupportVectorRegressionOptions Configuration options for Support Vector Regression (SVR), a powerful regression technique that uses support vector machines to predict continuous values.
SymbolicRegressionOptions Configuration options for Symbolic Regression, an evolutionary approach to finding mathematical expressions that best fit a dataset.
SyntheticDataGeneratorOptions<T> Options shared by the synthetic tabular data generators — the GAN, VAE and diffusion models in src/NeuralNetworks/SyntheticData that learn a data distribution and sample new rows from it.
TBATSModelOptions<T> Configuration options for the TBATS (Trigonometric seasonality, Box-Cox transformation, ARMA errors, Trend, and Seasonal components) time series forecasting model.
TCNOptions<T> Configuration options for the TCN (Temporal Convolutional Network) model.
TD3Options<T> Configuration options for TD3 agent.
TESTOptions<T> Configuration options for TEST (Text Embedding for Seasonality and Trend — Generating Text-Aligned Embeddings for Time Series).
TFCOptions<T> Configuration options for TF-C (Time-Frequency Consistency for Self-Supervised Time Series).
TOTEMOptions<T> Configuration options for TOTEM (TOkenized Time Series EMbeddings).
TOTOOptions<T> Configuration options for TOTO (Datadog’s Time Series Foundation Model for Observability).
TRPOOptions<T> Configuration options for Trust Region Policy Optimization (TRPO) agents.
TS2VecOptions<T> Configuration options for TS2Vec (Contrastive Learning of Universal Time Series Representations).
TSDiffOptions<T> Configuration options for TSDiff (Time Series Diffusion for unconditional/conditional generation).
TSMixerOptions<T> Configuration options for the TSMixer model.
TVAEOptions<T> Configuration options for TVAE (Tabular Variational Autoencoder), a VAE-based model for generating realistic synthetic tabular data.
TabDDPMOptions<T> Configuration options for TabDDPM (Tabular Denoising Diffusion Probabilistic Model), a diffusion-based model for generating realistic synthetic tabular data.
TabDPTOptions<T> Configuration options for TabDPT (Tabular Data Pre-Training).
TabFlowOptions<T> Configuration options for TabFlow, a flow matching model for generating synthetic tabular data using continuous normalizing flows with optimal transport paths.
TabLLMGenOptions<T> Configuration options for TabLLM-Gen, an LLM-style tabular data generator that uses schema-aware tokenization and autoregressive transformers with column prompts.
TabMOptions<T> Configuration options for TabM, a parameter-efficient ensemble model for tabular data.
TabNetOptions<T> Configuration options for TabNet, an attention-based interpretable deep learning model for tabular data.
TabPFNOptions<T> Configuration options for TabPFN (Prior-Fitted Networks for Tabular Data).
TabROptions<T> Configuration options for TabR, a retrieval-augmented model for tabular data.
TabSynOptions<T> Configuration options for TabSyn, a state-of-the-art synthetic tabular data generator that combines a VAE with latent diffusion for high-quality generation.
TabTransformerGenOptions<T> Configuration options for TabTransformer-Gen, a generative model that uses contextual embeddings from multi-head self-attention over categorical columns to generate realistic tabular data.
TabTransformerOptions<T> Configuration options for TabTransformer.
TableGANOptions<T> Configuration options for TableGAN, a DCGAN-style generative adversarial network for synthesizing tabular data with classification and information loss regularization.
TabuSearchOptions<T, TInput, TOutput> Configuration options for Tabu Search, a metaheuristic optimization algorithm that enhances local search by using memory structures to avoid revisiting previously explored solutions.
TabularActorCriticOptions<T> Configuration options for Tabular Actor-Critic agents.
TabularQLearningOptions<T> Configuration options for Tabular Q-Learning agents.
TeeAttestationOptions Configuration options for TEE remote attestation.
TeeOptions Configuration options for Trusted Execution Environment integration in federated learning.
TemporalFusionTransformerOptions<T> Configuration options for the Temporal Fusion Transformer (TFT) model.
TemporalGCNOptions<T> Configuration options for TemporalGCN (Temporal Graph Convolutional Network).
TextQueriedSegmentationOptions Options shared by segmentation models that classify pixels against text embeddings from a CLIP-style text encoder: the open-vocabulary models (CAT-Seg, SAN, SED, Mask-Adapter, Open-Vocabulary SAM) and the referring models (LISA, VideoLISA,…
ThompsonSamplingOptions<T> Configuration options for the Thompson-sampling bandit, which picks the arm whose sampled posterior reward is highest.
TiDEOptions<T> Options for TiDEModel — Time-series Dense Encoder (Das et al., “Long-term Forecasting with TiDE: Time-series Dense Encoder”, TMLR 2023).
TimeBridgeOptions<T> Configuration options for TimeBridge (Non-Stationarity Matters for Time Series Foundation Models).
TimeDiffOptions<T> Configuration options for TimeDiff (Non-autoregressive Diffusion-based Time Series Forecasting).
TimeGANOptions<T> Configuration options for TimeGAN, a generative adversarial network designed specifically for generating realistic time-series data while preserving temporal dynamics.
TimeGPTOptions<T> Configuration options for TimeGPT-style time series forecasting model.
TimeGradOptions<T> Configuration options for TimeGrad (Autoregressive Denoising Diffusion Model for Time Series).
TimeLLMOptions<T> Configuration options for Time-LLM (Large Language Model Reprogramming for Time Series).
TimeMAEOptions<T> Configuration options for TimeMAE (Masked Autoencoder for Time Series).
TimeMachineOptions<T> Configuration options for TimeMachine (Time Series State Space Model).
TimeMoEOptions<T> Configuration options for Time-MoE (Billion-Scale Time Series Foundation Models with Mixture of Experts).
TimeSeriesClassifierOptions<T> Configuration options for time series classifiers.
TimeSeriesCrossValidationFitDetectorOptions Configuration options for detecting overfitting, underfitting, and model stability in time series models using cross-validation techniques.
TimeSeriesFeatureOptions Configuration options for time series feature extraction transformers.
TimeSeriesForestOptions<T> Configuration options for the Time Series Forest classifier.
TimeSeriesIsolationForestOptions<T> Configuration options for Time Series Isolation Forest anomaly detection.
TimeSeriesRegressionOptions<T> Configuration options for time series regression models, which analyze data collected over time to identify patterns and make predictions.
TimerOptions<T> Configuration options for Timer (Generative Pre-Training for Time Series).
TimesFMOptions<T> Configuration options for the TimesFM (Time Series Foundation Model).
TimesNetOptions<T> Configuration options for the TimesNet model.
TinyTimeMixersOptions<T> Configuration options for Tiny Time Mixers (TTM) foundation model.
TradingAgentOptions<T> Configuration options for financial trading agents.
TrainingPipelineConfiguration<T, TInput, TOutput> Configuration for a multi-step training pipeline with customizable stages.
TransferFunctionOptions<T, TInput, TOutput> Configuration options for Transfer Function models, which model the dynamic relationship between input (exogenous) and output (endogenous) time series.
TrustRegionOptimizerOptions<T, TInput, TOutput> Configuration options for Trust Region optimization algorithms, which are robust methods for solving nonlinear optimization problems.
TweedieRegressionOptions<T> Configuration options for Tweedie Regression, a flexible generalized linear model that encompasses several distributions (Poisson, Gamma, Inverse Gaussian) as special cases.
UCBBanditOptions<T> Configuration options for the UCB bandit agent.
UncertaintyQuantificationOptions Configuration options for enabling uncertainty quantification during inference.
UniTSOptions<T> Configuration options for UniTS (Unified Time Series Model).
UnobservedComponentsOptions<T, TInput, TOutput> Configuration options for Unobserved Components Models (UCM), which decompose time series into trend, seasonal, cycle, and irregular components.
VARMAModelOptions<T> Configuration options for Vector Autoregressive Moving Average (VARMA) models, which extend VAR models by incorporating moving average terms.
VARModelOptions<T> Configuration options for Vector Autoregressive (VAR) models, which model the linear interdependencies among multiple time series.
VIFFitDetectorOptions Configuration options for detecting multicollinearity in regression models using Variance Inflation Factor (VIF) analysis.
VaROptions<T> Configuration options for Value-at-Risk (VaR) models.
ValueIterationOptions<T> Configuration options for Value Iteration agents.
VerificationOptions Configuration options for zero-knowledge verification in federated learning.
VerticalFederatedLearningOptions Top-level configuration for vertical federated learning (VFL).
VflUnlearningOptions Configuration for GDPR-compliant entity unlearning in vertical federated learning.
VisionTSOptions<T> Configuration options for VisionTS (Visual Masked Autoencoders as Zero-Shot Time Series Forecasters).
VoxelModelOptions Shared hyperparameters for the volumetric models that work on a 3D voxel grid.
WatkinsQLambdaOptions<T> Configuration options for Watkins’s Q(lambda), off-policy Q-learning with eligibility traces.
Wav2Vec2ModelOptions Configuration options for Wav2Vec2 speech recognition models.
WaveNetOptions<T> Configuration options for the WaveNet model adapted for time series forecasting.
WeightedRegressionOptions<T> Configuration options for weighted regression models, which assign different importance to different observations.
WorldModelsOptions<T> Configuration options for World Models agents (Ha & Schmidhuber 2018, “Recurrent World Models Facilitate Policy Evolution”, arXiv 1803.10122).
YingLongOptions<T> Configuration options for YingLong (Alibaba’s Enterprise Time Series Foundation Model).
ZeroInflatedRegressionOptions Configuration options for Zero-Inflated regression models.

Helpers & Utilities (7)

Type Summary
AiModelResultTrainingRecipeExtensions Reports what a trained model’s research paper specifies for training, what was actually used, and every difference between the two.
CloneEngine Executes a ClonePlan: reconstructs an instance and carries its configuration.
CloneRegistry Holds the compile-time clone plan for every discovered type, and builds one on demand for types the generator never saw.
ModelOptionsCloneExtensions Cloning for options classes.
ModelStateEnvelope Attaches declared state to a payload whose format this code does not know.
ParameterCollectionOrdering Deterministic collection ordering shared by generated model and network surfaces.
ParameterStableId Defines the canonical grammar used by durable parameter-manifest identities.

Exceptions (2)

Type Summary
ParameterContractViolationException Thrown when a parameter source’s concrete vector disagrees with the manifest snapshot that describes it.
ParameterLayoutNotReadyException Thrown when an exact vector operation is requested before every slot is ready.