RAGConfiguration class

Configuration for the RAG pipeline.

Specifies model paths, chunking parameters, and generation settings. Mirrors iOS RAGConfiguration exactly.

Constructors

RAGConfiguration({required String embeddingModelPath, required String llmModelPath, int embeddingDimension = 384, int topK = 3, double similarityThreshold = 0.3, int maxContextTokens = 2048, int chunkSize = 512, int chunkOverlap = 50, String? promptTemplate, String? embeddingConfigJSON, String? llmConfigJSON})
const

Properties

chunkOverlap int
Overlap between consecutive chunks in tokens (default: 50).
final
chunkSize int
Document chunk size in tokens (default: 512).
final
embeddingConfigJSON String?
Optional JSON configuration for the embedding model.
final
embeddingDimension int
Embedding vector dimension (default: 384).
final
embeddingModelPath String
Path to the ONNX embedding model file (required).
final
hashCode int
The hash code for this object.
no setterinherited
llmConfigJSON String?
Optional JSON configuration for the LLM.
final
llmModelPath String
Path to the GGUF LLM model file (required).
final
maxContextTokens int
Maximum context tokens to send to the LLM (default: 2048).
final
promptTemplate String?
Optional custom prompt template for the LLM.
final
runtimeType Type
A representation of the runtime type of the object.
no setterinherited
similarityThreshold double
Minimum cosine similarity threshold for retrieval (default: 0.3).
final
topK int
Number of top chunks to retrieve (default: 3).
final

Methods

noSuchMethod(Invocation invocation) → dynamic
Invoked when a nonexistent method or property is accessed.
inherited
toString() String
A string representation of this object.
override

Operators

operator ==(Object other) bool
The equality operator.
inherited