Class TokenSampling

Namespace

LMKit.TextGeneration.Sampling

Assembly

LM-Kit.NET.dll

Handles the sampling strategy used during text completion.

public abstract class TokenSampling

Inheritance

object

TokenSampling

Derived

GreedyDecoding

MirostatSampling

RandomSampling

TopNSigmaSampling

Inherited Members

object.Equals(object)

object.Equals(object, object)

object.GetHashCode()

object.GetType()

object.MemberwiseClone()

object.ReferenceEquals(object, object)

object.ToString()

Examples

The following example shows how to assign different sampling strategies to a conversation:

using LMKit.Model;
using LMKit.TextGeneration;
using LMKit.TextGeneration.Sampling;

LM model = LM.LoadFromModelID("gemma3:4b");
var chat = new MultiTurnConversation(model);

// Use random sampling with custom temperature.
chat.SamplingMode = new RandomSampling { Temperature = 0.7f };

// Switch to greedy decoding for deterministic output.
chat.SamplingMode = new GreedyDecoding();

// Use Mirostat v2 for perplexity-controlled generation.
chat.SamplingMode = new Mirostat2Sampling { TargetEntropy = 5.0f };

Methods

Clone()

Creates a deep copy of the current TokenSampling instance.