Class StackOverflowFederatedBenchmarkOptions

Namespace

AiDotNet.Configuration

Assembly

AiDotNet.dll

Configuration options for running the StackOverflow federated benchmark suite.

public sealed class StackOverflowFederatedBenchmarkOptions

Inheritance

object

StackOverflowFederatedBenchmarkOptions

Inherited Members

object.Equals(object)

object.Equals(object, object)

object.GetHashCode()

object.GetType()

object.ReferenceEquals(object, object)

object.ToString()

Remarks

StackOverflow is a large-scale federated text benchmark commonly used for next-token prediction. This v1 suite expects a LEAF-style JSON container with per-user token sequences.

For Beginners: You provide the train/test JSON files, and AiDotNet loads the per-user partitions, builds a vocabulary (with safe defaults), and evaluates your model on a standardized next-token task.

Properties

LoadOptions

Gets or sets load options controlling how many users/clients are loaded.

public LeafFederatedDatasetLoadOptions LoadOptions { get; set; }

Property Value

LeafFederatedDatasetLoadOptions

MaxSamplesPerUser

Gets or sets the maximum number of samples to use per user/client (null uses all available).

public int? MaxSamplesPerUser { get; set; }

Property Value

int?

MaxVocabularySize

Gets or sets the maximum vocabulary size used for token-to-ID mapping.

public int? MaxVocabularySize { get; set; }

Property Value

int?

Remarks

If null, AiDotNet uses a sensible default (smaller in CI mode).

SequenceLength

Gets or sets the fixed token sequence length used as model input.

public int? SequenceLength { get; set; }

Property Value

int?

Remarks

If null, AiDotNet infers the length from the dataset and validates consistency.

TestFilePath

Gets or sets the optional path to the StackOverflow test split JSON file.

public string? TestFilePath { get; set; }

Property Value

string

TrainFilePath

Gets or sets the path to the StackOverflow train split JSON file.

public string? TrainFilePath { get; set; }

Property Value

string

VocabularyTrainingSampleCount

Gets or sets the maximum number of sequences used to build the default vocabulary.

public int? VocabularyTrainingSampleCount { get; set; }

Property Value

int?

Remarks

If null, AiDotNet uses a sensible default (smaller in CI mode).