Expand description
LLM integration
Provides traits and implementations for:
- Embedding generation via external services (vLLM, OpenAI, etc.)
- Document metadata generation
- Query parsing
- Reranking
Structs§
- Chat
Message - Chat message for completion requests
- Document
Metadata - Generated metadata result
- Expanded
Query - Expanded query variants
- Http
Embedder - Embedder that uses external HTTP service (vLLM, OpenAI, etc.)
- Http
Metadata Generator - Metadata generator using external HTTP LLM service
- Http
Query Parser - Query parser using external HTTP LLM service
- Llama
Embedder - LLaMA-based embedder
- Llama
Metadata Generator - LLaMA-based metadata generator
- Metadata
Context - Context information for metadata generation
- Metadata
Filter Hint - Metadata filter hint extracted from query
- Metrics
Snapshot - Snapshot of API metrics
- Parsed
Query - Parsed query with extracted intent and filters
- Query
Parser - Query parser using local LLM
- Rerank
Document - Document for reranking
- Rerank
Result - Reranking result
- Temporal
Filter - Temporal filter for time-based queries
- VLLM
Client - vLLM/OpenAI-compatible client
Enums§
- Search
Type - Search type recommendation
Constants§
- DEFAULT_
EMBED_ MODEL - Default embedding model (nomic-embed-text or similar)
- DEFAULT_
METADATA_ MODEL - Default metadata generation model
Traits§
- Embedder
- Embedding generation trait
- LLMClient
- Trait for LLM service clients
- Metadata
Generator - Metadata generation trait
- Query
Expander - Query expansion trait
- Reranker
- Document reranking trait
- Tokenizer
- Tokenization trait
Functions§
- generate_
metadata_ with_ llm - Helper to generate metadata using LLM client