Configuration defaults

Default values for all GraphRAG configuration options

GraphRAG provides sensible defaults for all configuration options. This reference documents all default values.

Model defaults

Completion models

DEFAULT_COMPLETION_MODEL_ID · str · default: "default_completion_model"

Default identifier for completion model configurations.

DEFAULT_COMPLETION_MODEL · str · default: "gpt-4.1"

Default completion model name.

DEFAULT_COMPLETION_MODEL_AUTH_TYPE · AuthMethod · default: "ApiKey"

Default authentication method for completion models.

DEFAULT_MODEL_PROVIDER · str · default: "openai"

Default model provider.

Embedding models

DEFAULT_EMBEDDING_MODEL_ID · str · default: "default_embedding_model"

Default identifier for embedding model configurations.

DEFAULT_EMBEDDING_MODEL · str · default: "text-embedding-3-large"

Default embedding model name.

DEFAULT_EMBEDDING_MODEL_AUTH_TYPE · AuthMethod · default: "ApiKey"

Default authentication method for embedding models.

Encoding

ENCODING_MODEL · str · default: "o200k_base"

Default encoding model for tokenization.

Directory defaults

DEFAULT_INPUT_BASE_DIR · str · default: "input"

Default base directory for input files.

DEFAULT_OUTPUT_BASE_DIR · str · default: "output"

Default base directory for output files.

DEFAULT_CACHE_BASE_DIR · str · default: "cache"

Default base directory for cache storage.

DEFAULT_UPDATE_OUTPUT_BASE_DIR · str · default: "update_output"

Default base directory for incremental update output.

Entity types

DEFAULT_ENTITY_TYPES · list[str]

Default entity types to extract during graph construction.

Configuration class defaults

The following sections document default values for each configuration class.

BasicSearchDefaults

prompt · None · default: "None"

Basic search prompt template.

k · int · default: "10"

Number of results to return.

max_context_tokens · int · default: "12000"

Maximum context tokens.

completion_model_id · str · default: "default_completion_model"

Completion model ID.

embedding_model_id · str · default: "default_embedding_model"

Embedding model ID.

ChunkingDefaults

type · str · default: "tokens"

Chunking strategy type (from ChunkerType enum).

size · int · default: "1200"

Chunk size in tokens.

overlap · int · default: "100"

Overlap between chunks in tokens.

encoding_model · str · default: "o200k_base"

Encoding model for tokenization.

prepend_metadata · None · default: "None"

Metadata to prepend to chunks.

ClusterGraphDefaults

max_cluster_size · int · default: "10"

Maximum size of clusters.

use_lcc · bool · default: "True"

Whether to use the largest connected component.

seed · int · default: "0xDEADBEEF"

Random seed for clustering (3735928559 in decimal).

CommunityReportDefaults

graph_prompt · None · default: "None"

Prompt for graph-based community reports.

text_prompt · None · default: "None"

Prompt for text-based community reports.

max_length · int · default: "2000"

Maximum report length in tokens.

max_input_length · int · default: "8000"

Maximum input length in tokens.

completion_model_id · str · default: "default_completion_model"

Completion model ID.

model_instance_name · str · default: "community_reporting"

Model instance name for caching.

DriftSearchDefaults

prompt · None · default: "None"

DRIFT search prompt.

reduce_prompt · None · default: "None"

Reduce step prompt.

data_max_tokens · int · default: "12000"

Maximum data tokens.

reduce_max_tokens · None · default: "None"

Maximum reduce tokens.

reduce_temperature · float · default: "0"

Temperature for reduce step.

reduce_max_completion_tokens · None · default: "None"

Maximum completion tokens for reduce.

concurrency · int · default: "32"

Concurrency level for DRIFT operations.

drift_k_followups · int · default: "20"

Number of followup queries.

primer_folds · int · default: "5"

Number of primer folds.

primer_llm_max_tokens · int · default: "12000"

Maximum tokens for primer LLM.

n_depth · int · default: "3"

Search depth.

local_search_text_unit_prop · float · default: "0.9"

Text unit proportion for local search component.

local_search_community_prop · float · default: "0.1"

Community proportion for local search component.

local_search_top_k_mapped_entities · int · default: "10"

Top k entities for local search.

local_search_top_k_relationships · int · default: "10"

Top k relationships for local search.

local_search_max_data_tokens · int · default: "12000"

Maximum data tokens for local search.

local_search_temperature · float · default: "0"

Temperature for local search.

local_search_top_p · float · default: "1"

Top p for local search.

local_search_n · int · default: "1"

Number of completions for local search.

local_search_llm_max_gen_tokens · int | None · default: "None"

Maximum generation tokens for local search.

local_search_llm_max_gen_completion_tokens · int | None · default: "None"

Maximum completion tokens for local search.

completion_model_id · str · default: "default_completion_model"

Completion model ID.

embedding_model_id · str · default: "default_embedding_model"

Embedding model ID.

EmbedTextDefaults

embedding_model_id · str · default: "default_embedding_model"

Embedding model ID.

model_instance_name · str · default: "text_embedding"

Model instance name for caching.

batch_size · int · default: "16"

Batch size for embedding operations.

batch_max_tokens · int · default: "8191"

Maximum tokens per batch.

names · list[str]

List of embeddings to generate (uses default_embeddings).

ExtractClaimsDefaults

enabled · bool · default: "False"

Whether claim extraction is enabled.

prompt · None · default: "None"

Claim extraction prompt.

description · str

Description of claims to extract.

max_gleanings · int · default: "1"

Maximum number of gleaning iterations.

completion_model_id · str · default: "default_completion_model"

Completion model ID.

model_instance_name · str · default: "extract_claims"

Model instance name for caching.

ExtractGraphDefaults

prompt · None · default: "None"

Graph extraction prompt.

entity_types · list[str]

Entity types to extract.

max_gleanings · int · default: "1"

Maximum number of gleaning iterations.

completion_model_id · str · default: "default_completion_model"

Completion model ID.

model_instance_name · str · default: "extract_graph"

Model instance name for caching.

TextAnalyzerDefaults

Used for NLP-based graph extraction.

extractor_type · NounPhraseExtractorType · default: "RegexEnglish"

Noun phrase extractor type.

model_name · str · default: "en_core_web_md"

SpaCy model name.

max_word_length · int · default: "15"

Maximum word length to consider.

word_delimiter · str · default: "' '"

Delimiter between words.

include_named_entities · bool · default: "True"

Whether to include named entities.

exclude_nouns · list[str]

List of nouns to exclude (uses EN_STOP_WORDS).

exclude_entity_tags · list[str] · default: "[\"DATE\"]"

Entity tags to exclude.

exclude_pos_tags · list[str] · default: "[\"DET\", \"PRON\", \"INTJ\", \"X\"]"

Part-of-speech tags to exclude.

noun_phrase_tags · list[str] · default: "[\"PROPN\", \"NOUNS\"]"

Tags for noun phrases.

noun_phrase_grammars · dict[str, str]

Grammar rules for noun phrase combination.

Default:

{
    "PROPN,PROPN": "PROPN",
    "NOUN,NOUN": "NOUNS",
    "NOUNS,NOUN": "NOUNS",
    "ADJ,ADJ": "ADJ",
    "ADJ,NOUN": "NOUNS"
}

ExtractGraphNLPDefaults

normalize_edge_weights · bool · default: "True"

Whether to normalize edge weights.

text_analyzer · TextAnalyzerDefaults

Text analyzer configuration.

concurrent_requests · int · default: "25"

Number of concurrent requests.

async_mode · AsyncType · default: "Threaded"

Async mode to use.

GlobalSearchDefaults

map_prompt · None · default: "None"

Map step prompt.

reduce_prompt · None · default: "None"

Reduce step prompt.

knowledge_prompt · None · default: "None"

Knowledge generation prompt.

max_context_tokens · int · default: "12000"

Maximum context tokens.

data_max_tokens · int · default: "12000"

Maximum data tokens.

map_max_length · int · default: "1000"

Maximum map response length in words.

reduce_max_length · int · default: "2000"

Maximum reduce response length in words.

dynamic_search_threshold · int · default: "1"

Community rating threshold for inclusion.

dynamic_search_keep_parent · bool · default: "False"

Keep parent community if children are relevant.

dynamic_search_num_repeats · int · default: "1"

Number of times to rate each community.

dynamic_search_use_summary · bool · default: "False"

Use community summary instead of full context.

dynamic_search_max_level · int · default: "2"

Maximum community hierarchy level.

completion_model_id · str · default: "default_completion_model"

Completion model ID.

StorageDefaults

type · str · default: "file"

Storage type (from StorageType enum).

encoding · str | None · default: "None"

Text encoding.

base_dir · str | None · default: "None"

Base directory for file storage.

azure_connection_string · None · default: "None"

Azure connection string.

azure_container_name · None · default: "None"

Azure container name.

azure_account_url · None · default: "None"

Azure account URL.

azure_cosmosdb_account_url · None · default: "None"

Azure CosmosDB account URL.

InputDefaults

type · InputType · default: "Text"

Input type.

encoding · str | None · default: "None"

Text encoding.

file_pattern · None · default: "None"

File pattern for matching input files.

id_column · None · default: "None"

Column name for document IDs.

title_column · None · default: "None"

Column name for document titles.

text_column · None · default: "None"

Column name for document text.

InputStorageDefaults

Extends StorageDefaults.

base_dir · str · default: "input"

Base directory for input storage.

CacheStorageDefaults

Extends StorageDefaults.

base_dir · str · default: "cache"

Base directory for cache storage.

CacheDefaults

type · CacheType · default: "Json"

Cache type.

storage · CacheStorageDefaults

Cache storage configuration.

LocalSearchDefaults

prompt · None · default: "None"

Local search prompt.

text_unit_prop · float · default: "0.5"

Text unit proportion.

community_prop · float · default: "0.15"

Community proportion.

conversation_history_max_turns · int · default: "5"

Maximum conversation history turns.

top_k_entities · int · default: "10"

Top k entities to retrieve.

top_k_relationships · int · default: "10"

Top k relationships to retrieve.

max_context_tokens · int · default: "12000"

Maximum context tokens.

completion_model_id · str · default: "default_completion_model"

Completion model ID.

embedding_model_id · str · default: "default_embedding_model"

Embedding model ID.

OutputStorageDefaults

Extends StorageDefaults.

base_dir · str · default: "output"

Base directory for output storage.

PruneGraphDefaults

min_node_freq · int · default: "2"

Minimum node frequency.

max_node_freq_std · None · default: "None"

Maximum node frequency standard deviation.

min_node_degree · int · default: "1"

Minimum node degree.

max_node_degree_std · None · default: "None"

Maximum node degree standard deviation.

min_edge_weight_pct · float · default: "40.0"

Minimum edge weight percentage.

remove_ego_nodes · bool · default: "True"

Whether to remove ego nodes.

lcc_only · bool · default: "False"

Keep only largest connected component.

ReportingDefaults

type · ReportingType · default: "file"

Reporting type.

base_dir · str · default: "logs"

Base directory for reporting.

connection_string · None · default: "None"

Connection string for blob reporting.

container_name · None · default: "None"

Container name for blob reporting.

storage_account_blob_url · None · default: "None"

Storage account blob URL.

SnapshotsDefaults

embeddings · bool · default: "False"

Whether to save embedding snapshots.

graphml · bool · default: "False"

Whether to save GraphML snapshots.

raw_graph · bool · default: "False"

Whether to save raw graph snapshots.

SummarizeDescriptionsDefaults

prompt · None · default: "None"

Summarization prompt.

max_length · int · default: "500"

Maximum summary length in tokens.

max_input_tokens · int · default: "4000"

Maximum input tokens.

completion_model_id · str · default: "default_completion_model"

Completion model ID.

model_instance_name · str · default: "summarize_descriptions"

Model instance name for caching.

UpdateOutputStorageDefaults

Extends StorageDefaults.

base_dir · str · default: "update_output"

Base directory for update output storage.

VectorStoreDefaults

type · str · default: "lancedb"

Vector store type (from VectorStoreType enum).

db_uri · str · default: "output/lancedb"

Database URI for vector store.

GraphRagConfigDefaults

Root configuration defaults.

models · dict · default: "{}"

Legacy model configurations.

completion_models · dict · default: "{}"

Completion model configurations.

embedding_models · dict · default: "{}"

Embedding model configurations.

concurrent_requests · int · default: "25"

Default concurrent requests.

async_mode · AsyncType · default: "Threaded"

Default async mode.

reporting · ReportingDefaults

Reporting configuration defaults.

input_storage · InputStorageDefaults

Input storage configuration defaults.

output_storage · OutputStorageDefaults

Output storage configuration defaults.

update_output_storage · UpdateOutputStorageDefaults

Update output storage configuration defaults.

cache · CacheDefaults

Cache configuration defaults.

input · InputDefaults

Input configuration defaults.

embed_text · EmbedTextDefaults

Text embedding configuration defaults.

chunking · ChunkingDefaults

Chunking configuration defaults.

snapshots · SnapshotsDefaults

Snapshots configuration defaults.

extract_graph · ExtractGraphDefaults

Graph extraction configuration defaults.

extract_graph_nlp · ExtractGraphNLPDefaults

NLP graph extraction configuration defaults.

summarize_descriptions · SummarizeDescriptionsDefaults

Description summarization configuration defaults.

community_reports · CommunityReportDefaults

Community reports configuration defaults.

extract_claims · ExtractClaimsDefaults

Claims extraction configuration defaults.

prune_graph · PruneGraphDefaults

Graph pruning configuration defaults.

cluster_graph · ClusterGraphDefaults

Graph clustering configuration defaults.

local_search · LocalSearchDefaults

Local search configuration defaults.

global_search · GlobalSearchDefaults

Global search configuration defaults.

drift_search · DriftSearchDefaults

DRIFT search configuration defaults.

basic_search · BasicSearchDefaults

Basic search configuration defaults.

vector_store · VectorStoreDefaults

Vector store configuration defaults.

workflows · None · default: "None"

Workflows list.