Class ReinforcementTuningHyperParameters
- Namespace
- GenerativeAI.Types
- Assembly
- GenerativeAI.dll
Hyperparameters for reinforcement tuning.
[JsonNumberHandling(JsonNumberHandling.AllowReadingFromString)]
public class ReinforcementTuningHyperParameters
- Inheritance
-
ReinforcementTuningHyperParameters
- Inherited Members
- Extension Methods
Properties
AdapterSize
Optional. Adapter size for reinforcement tuning.
[JsonPropertyName("adapterSize")]
public AdapterSize? AdapterSize { get; set; }
Property Value
BatchSize
Optional. Batch size (number of prompts processed per train step).
[JsonPropertyName("batchSize")]
public int? BatchSize { get; set; }
Property Value
- int?
CheckpointInterval
Optional. How often, in steps, to save checkpoints during training.
[JsonPropertyName("checkpointInterval")]
public int? CheckpointInterval { get; set; }
Property Value
- int?
EpochCount
Optional. Number of training epochs for the tuning job.
[JsonPropertyName("epochCount")]
[JsonNumberHandling(JsonNumberHandling.AllowReadingFromString)]
public long? EpochCount { get; set; }
Property Value
- long?
EvaluateInterval
Optional. How often, in steps, to evaluate the tuning job during training.
[JsonPropertyName("evaluateInterval")]
public int? EvaluateInterval { get; set; }
Property Value
- int?
LearningRateMultiplier
Learning rate multiplier for reinforcement learning.
[JsonPropertyName("learningRateMultiplier")]
public double? LearningRateMultiplier { get; set; }
Property Value
MaxOutputTokens
Optional. The maximum number of tokens to generate per prompt.
[JsonPropertyName("maxOutputTokens")]
public int? MaxOutputTokens { get; set; }
Property Value
- int?
SamplesPerPrompt
Optional. Number of different responses to generate per prompt during tuning.
[JsonPropertyName("samplesPerPrompt")]
public int? SamplesPerPrompt { get; set; }
Property Value
- int?
StepCount
Optional. Number of steps for the tuning job (mutually exclusive with EpochCount).
[JsonPropertyName("stepCount")]
[JsonNumberHandling(JsonNumberHandling.AllowReadingFromString)]
public long? StepCount { get; set; }
Property Value
- long?
ThinkingBudget
Optional. The thinking budget to optimize for (Gemini 2.5 only; -1 means dynamic thinking).
[JsonPropertyName("thinkingBudget")]
public int? ThinkingBudget { get; set; }
Property Value
- int?
ThinkingLevel
Optional. The maximum thinking depth during tuning.
[JsonPropertyName("thinkingLevel")]
public ReinforcementTuningThinkingLevel? ThinkingLevel { get; set; }