59e89772de
Remove legacy storage implementations and deprecated examples: - Delete FAISS, JSON, Memgraph, Milvus, MongoDB, Nano Vector DB, Neo4j, NetworkX, Qdrant, Redis storage backends - Remove Kubernetes deployment manifests and installation scripts - Delete unofficial examples for deprecated backends and offline deployment docs Streamline core infrastructure: - Consolidate storage layer to PostgreSQL-only implementation - Add full-text search caching with FTS cache module - Implement metrics collection and monitoring pipeline - Add explain and metrics API routes Modernize frontend and tooling: - Switch web UI to Bun with bun.lock, remove npm and pnpm lockfiles - Update Dockerfile for PostgreSQL-only deployment - Add Makefile for common development tasks - Update environment and configuration examples Enhance evaluation and testing capabilities: - Add prompt optimization with DSPy and auto-tuning - Implement ground truth regeneration and variant testing - Add prompt debugging and response comparison utilities - Expand test coverage with new integration scenarios Simplify dependencies and configuration: - Remove offline-specific requirement files - Update pyproject.toml with streamlined dependencies - Add Python version pinning with .python-version - Create project guidelines in CLAUDE.md and AGENTS.md
140 lines
5.6 KiB
Python
140 lines
5.6 KiB
Python
"""
|
|
Centralized configuration constants for LightRAG.
|
|
|
|
This module defines default values for configuration constants used across
|
|
different parts of the LightRAG system. Centralizing these values ensures
|
|
consistency and makes maintenance easier.
|
|
"""
|
|
|
|
# Default values for server settings
|
|
DEFAULT_WOKERS = 2
|
|
DEFAULT_MAX_GRAPH_NODES = 1000
|
|
|
|
# Default values for extraction settings
|
|
DEFAULT_SUMMARY_LANGUAGE = 'English' # Default language for document processing
|
|
DEFAULT_MAX_GLEANING = 1
|
|
DEFAULT_ENTITY_NAME_MAX_LENGTH = 256
|
|
|
|
# Number of description fragments to trigger LLM summary
|
|
DEFAULT_FORCE_LLM_SUMMARY_ON_MERGE = 8
|
|
# Max description token size to trigger LLM summary
|
|
DEFAULT_SUMMARY_MAX_TOKENS = 1200
|
|
# Recommended LLM summary output length in tokens
|
|
DEFAULT_SUMMARY_LENGTH_RECOMMENDED = 600
|
|
# Maximum token size sent to LLM for summary
|
|
DEFAULT_SUMMARY_CONTEXT_SIZE = 12000
|
|
# Default entities to extract if ENTITY_TYPES is not specified in .env
|
|
DEFAULT_ENTITY_TYPES = [
|
|
'Person',
|
|
'Creature',
|
|
'Organization',
|
|
'Location',
|
|
'Event',
|
|
'Concept',
|
|
'Method',
|
|
'Content',
|
|
'Data',
|
|
'Artifact',
|
|
'NaturalObject',
|
|
]
|
|
|
|
# Separator for: description, source_id and relation-key fields(Can not be changed after data inserted)
|
|
GRAPH_FIELD_SEP = '<SEP>'
|
|
|
|
# Query and retrieval configuration defaults
|
|
DEFAULT_TOP_K = 60
|
|
DEFAULT_CHUNK_TOP_K = 40 # Increased from 30 to improve context recall
|
|
# Token limits increased for modern 128K+ context models (GPT-4o-mini, Claude, etc.)
|
|
# Top-k already limits count; these are safety ceilings, not aggressive truncation targets
|
|
DEFAULT_MAX_ENTITY_TOKENS = 16000
|
|
DEFAULT_MAX_RELATION_TOKENS = 16000
|
|
DEFAULT_MAX_TOTAL_TOKENS = 60000
|
|
DEFAULT_COSINE_THRESHOLD = 0.30 # Lowered from 0.35 to retrieve more related content
|
|
DEFAULT_RELATED_CHUNK_NUMBER = 16 # Increased from 12 to improve context recall
|
|
DEFAULT_KG_CHUNK_PICK_METHOD = 'VECTOR'
|
|
|
|
# TODO: Deprated. All conversation_history messages is send to LLM.
|
|
DEFAULT_HISTORY_TURNS = 0
|
|
|
|
# Rerank configuration defaults
|
|
# Local reranking uses mxbai-rerank-xsmall-v1 by default (see rerank.py)
|
|
DEFAULT_ENABLE_RERANK = True
|
|
# Minimum rerank score threshold - set to None to disable filtering
|
|
# Testing shows reranking works best for ordering only, without score cutoffs
|
|
# (filtering hurts recall on domain-specific content)
|
|
DEFAULT_MIN_RERANK_SCORE = None
|
|
|
|
# Two-stage retrieval configuration
|
|
# When reranking is enabled, retrieve more candidates than chunk_top_k to surface
|
|
# hidden relevant chunks. The reranker then selects the best chunk_top_k results.
|
|
# Example: chunk_top_k=40, multiplier=2 → retrieve 80 → rerank → return top 40
|
|
# Note: Higher multipliers (2-3x) help when recall is low but can hurt precision
|
|
# if the reranker isn't accurate enough. Set to 1 to disable two-stage retrieval.
|
|
# Testing shows 1x (disabled) works best when baseline recall is already high (>95%).
|
|
DEFAULT_RETRIEVAL_MULTIPLIER = 1 # 1 = disabled, 2-3 = enabled
|
|
|
|
# Default source ids limit in meta data for entity and relation
|
|
DEFAULT_MAX_SOURCE_IDS_PER_ENTITY = 300
|
|
DEFAULT_MAX_SOURCE_IDS_PER_RELATION = 300
|
|
### control chunk_ids limitation method: KEEP, FIFO
|
|
### KEEP: Keep oldest (less merge action and faster)
|
|
### FIFO: First in first out
|
|
SOURCE_IDS_LIMIT_METHOD_KEEP = 'KEEP'
|
|
SOURCE_IDS_LIMIT_METHOD_FIFO = 'FIFO'
|
|
DEFAULT_SOURCE_IDS_LIMIT_METHOD = SOURCE_IDS_LIMIT_METHOD_FIFO
|
|
VALID_SOURCE_IDS_LIMIT_METHODS = {
|
|
SOURCE_IDS_LIMIT_METHOD_KEEP,
|
|
SOURCE_IDS_LIMIT_METHOD_FIFO,
|
|
}
|
|
# Maximum number of file paths stored in entity/relation file_path field (For displayed only, does not affect query performance)
|
|
DEFAULT_MAX_FILE_PATHS = 100
|
|
|
|
# Placeholder when file_path list exceeds DEFAULT_MAX_FILE_PATHS (used by all storage backends)
|
|
DEFAULT_FILE_PATH_MORE_PLACEHOLDER = 'truncated'
|
|
|
|
# Default temperature for LLM (lower = more deterministic, less hallucination risk)
|
|
# Using 0.1 for evaluation stability; production may use 0.3-0.7 for more varied responses
|
|
DEFAULT_TEMPERATURE = 0.1
|
|
|
|
# Async configuration defaults
|
|
DEFAULT_MAX_ASYNC = 4 # Default maximum async operations
|
|
DEFAULT_MAX_PARALLEL_INSERT = 2 # Default maximum parallel insert operations
|
|
|
|
# Embedding configuration defaults
|
|
DEFAULT_EMBEDDING_FUNC_MAX_ASYNC = 8 # Default max async for embedding functions
|
|
DEFAULT_EMBEDDING_BATCH_NUM = 10 # Default batch size for embedding computations
|
|
|
|
# Gunicorn worker timeout
|
|
DEFAULT_TIMEOUT = 300
|
|
|
|
# Default llm and embedding timeout
|
|
DEFAULT_LLM_TIMEOUT = 180
|
|
DEFAULT_EMBEDDING_TIMEOUT = 30
|
|
|
|
# Topic connectivity check configuration
|
|
DEFAULT_MIN_RELATIONSHIP_DENSITY = 0.3 # Minimum ratio of relationships to entities
|
|
DEFAULT_MIN_ENTITY_COVERAGE = 0.5 # Minimum ratio of entities connected by relationships
|
|
DEFAULT_CHECK_TOPIC_CONNECTIVITY = True # Enable topic connectivity check by default
|
|
|
|
# Logging configuration defaults
|
|
DEFAULT_LOG_MAX_BYTES = 10485760 # Default 10MB
|
|
DEFAULT_LOG_BACKUP_COUNT = 5 # Default 5 backups
|
|
DEFAULT_LOG_FILENAME = 'lightrag.log' # Default log filename
|
|
|
|
# Ollama server configuration defaults
|
|
DEFAULT_OLLAMA_MODEL_NAME = 'lightrag'
|
|
DEFAULT_OLLAMA_MODEL_TAG = 'latest'
|
|
DEFAULT_OLLAMA_MODEL_SIZE = 7365960935
|
|
DEFAULT_OLLAMA_CREATED_AT = '2024-01-15T00:00:00Z'
|
|
DEFAULT_OLLAMA_DIGEST = 'sha256:lightrag'
|
|
|
|
# Full-text search cache configuration
|
|
# Shorter TTL than embedding cache since document content changes more frequently
|
|
DEFAULT_FTS_CACHE_TTL = 300 # 5 minutes
|
|
DEFAULT_FTS_CACHE_MAX_SIZE = 5000 # Smaller than embedding cache
|
|
DEFAULT_FTS_CACHE_ENABLED = True
|
|
|
|
# Metrics configuration
|
|
DEFAULT_METRICS_ENABLED = True
|
|
DEFAULT_METRICS_HISTORY_SIZE = 1000 # Queries to keep in circular buffer
|
|
DEFAULT_METRICS_WINDOW_SECONDS = 3600 # 1 hour window for percentile calculations
|