---
title: SampleGenerator
framework: evaluations
role: symbol
role_heading: Class
path: evaluations/samplegenerator
---

# SampleGenerator

An actor that generates evaluation samples using a language model.

## Declaration

```swift
actor SampleGenerator<SampleType> where SampleType : ModelSampleProtocol
```

## Mentioned in

Designing datasets to test your feature Generating synthetic datasets Designing effective evaluations

## Overview

Overview Create a generator, configure its properties, then call run() to produce new samples as an async stream. After iteration completes, access samples for all generated samples, or invalidSamples for any the validator rejected.

## Topics

### Configuring generation

- [samplingStrategy](evaluations/samplegenerator/samplingstrategy-swift.property.md)
- [validator](evaluations/samplegenerator/validator.md)
- [SampleGenerator.SamplingStrategy](evaluations/samplegenerator/samplingstrategy-swift.enum.md)

### Running generation

- [run()](evaluations/samplegenerator/run().md)

### Accessing results

- [samples](evaluations/samplegenerator/samples.md)
- [invalidSamples](evaluations/samplegenerator/invalidsamples.md)

### Initializers

- [init(_:samples:targetCount:sessionProvider:samplingStrategy:validator:)](evaluations/samplegenerator/init(_:samples:targetcount:sessionprovider:samplingstrategy:validator:)-2804x.md)
- [init(_:samples:targetCount:sessionProvider:samplingStrategy:validator:)](evaluations/samplegenerator/init(_:samples:targetcount:sessionprovider:samplingstrategy:validator:)-8t01x.md)

## Relationships

### Conforms To

- [Actor](swift/actor.md)
- [Sendable](swift/sendable.md)
- [SendableMetatype](swift/sendablemetatype.md)

## See Also

### Datasets

- [Generating synthetic datasets](evaluations/generating-synthetic-evaluation-datasets.md)
- [Designing datasets to test your feature](evaluations/designing-evaluation-datasets.md)
- [ModelSample](evaluations/modelsample.md)
- [Loader](evaluations/loader.md)