Method AddSample

Namespace

LMKit.TextAnalysis.Training

Assembly

LM-Kit.NET.dll

AddSample(string, IList<string>, IList<string>, int)

Adds a training sample from raw text using the engine’s preferred modality.

public void AddSample(string content, IList<string> categories, IList<string> categoryDescriptions, int resultIndex)

Parameters

content string

The textual content to classify.

categories IList<string>

The list of category labels. The item at resultIndex is treated as the correct label for this sample.

categoryDescriptions IList<string>

Optional human-readable descriptions aligned 1:1 with categories.

resultIndex int

The zero-based index into categories indicating the expected class.

Examples

dataset.AddSample(
    "The battery lasts two days on a single charge.",
    new[] { "Positive", "Neutral", "Negative" },
    new[] { "Favorable tone", "Objective tone", "Unfavorable tone" },
    resultIndex: 0);

Remarks

The content is wrapped as a text Attachment and forwarded to AddSample(InferenceModality, Attachment, IList<string>, IList<string>, int).

AddSample(Attachment, IList<string>, IList<string>, int)

Adds a training sample from an Attachment using the engine’s preferred modality.

public void AddSample(Attachment content, IList<string> categories, IList<string> categoryDescriptions, int resultIndex)

Parameters

content Attachment

The input attachment (e.g., text, image, or multimodal source) to classify.

categories IList<string>

The list of category labels. The item at resultIndex is treated as the correct label for this sample.

categoryDescriptions IList<string>

Optional human-readable descriptions aligned 1:1 with categories.

resultIndex int

The zero-based index into categories indicating the expected class.

Remarks

For text inputs, prefer CreateFromText(string, string). This overload delegates to the modality-explicit overload.

AddSample(InferenceModality, Attachment, IList<string>, IList<string>, int)

Adds a training sample with an explicit InferenceModality.

public void AddSample(InferenceModality modality, Attachment content, IList<string> categories, IList<string> categoryDescriptions, int resultIndex)

Parameters

modality InferenceModality

The inference modality to use for prompt generation.

content Attachment

The content attachment to classify.

categories IList<string>

The list of category labels. The item at resultIndex is the expected label for this sample.

categoryDescriptions IList<string>

Optional human-readable descriptions aligned 1:1 with categories.

resultIndex int

The zero-based index of the correct label within categories.

Examples

var attachment = Attachment.CreateFromText(reviewText, "text");
var labels = new[] { "Positive", "Neutral", "Negative" };
var desc   = new[] { "Favorable tone", "Objective tone", "Unfavorable tone" };

dataset.AddSample(
    InferenceModality.Text,
    attachment,
    labels,
    desc,
    resultIndex: 2);

Remarks

This method calls Categorization.CreateSamplingClassificationTrainingSample to synthesize a chat-based training example using Classification.CategorizationEngine.NormalizationLevel.All. The resulting sample is wrapped in ChatTrainingSample with the resolved inferenceModality.

If EnableModalityAugmentation is true and the resolved inferenceModality is Multimodal, two additional samples are automatically added using Text and Vision.

The lengths of categories and categoryDescriptions should match (when descriptions are provided), and resultIndex must be a valid index into categories.