ProgramSynthesisOptions
Options & Configuration · AiDotNet.ProgramSynthesis.Options
Configuration options for enabling Program Synthesis / Code Tasks via the primary builder/result APIs.
For Beginners
These settings control which built-in code model to use and how big it is. You can usually accept the defaults and only change the language and model kind.
How It Works
This options type is intended for use with AiModelBuilder to configure a code model and
safe defaults (tokenization and architecture) without requiring users to construct low-level engines.
Constructors
| Constructor | Summary |
|---|---|
ProgramSynthesisOptions() |
Properties
| Property | Summary |
|---|---|
CodeTask DefaultTask { get; set; } |
Gets or sets the default code task type used for the configured architecture. |
ICodeSyntaxProvider SyntaxProvider { get; set; } |
Gets or sets the parser used to validate generated code and to extract syntax trees, for example TreeSitterCodeSyntaxProvider from the opt-in AiDotNet.CodeAnalysis.TreeSitter package. |
ITokenizer Tokenizer { get; set; } |
Gets or sets an optional tokenizer to use for code tasks. |
ProgramLanguage TargetLanguage { get; set; } |
Gets or sets the target programming language for the configured model. |
ProgramSynthesisModelKind ModelKind { get; set; } |
Gets or sets the built-in code model kind to configure when no explicit model is provided. |
SynthesisType SynthesisType { get; set; } |
Gets or sets the synthesis approach type used for the configured architecture. |
int MaxSequenceLength { get; set; } |
Gets or sets the maximum sequence length (in tokens) that the model can process. |
int NumDecoderLayers { get; set; } |
Gets or sets the number of decoder layers used by the configured architecture. |
int NumEncoderLayers { get; set; } |
Gets or sets the number of encoder layers used by the configured architecture. |
int VocabularySize { get; set; } |
Gets or sets the vocabulary size used by the configured architecture. |