> ## Documentation Index
> Fetch the complete documentation index at: https://mintlify.com/microsoft/graphrag/llms.txt
> Use this file to discover all available pages before exploring further.
# 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 identifier for completion model configurations.
Default completion model name.
Default authentication method for completion models.
Default model provider.
### Embedding models
Default identifier for embedding model configurations.
Default embedding model name.
Default authentication method for embedding models.
### Encoding
Default encoding model for tokenization.
## Directory defaults
Default base directory for input files.
Default base directory for output files.
Default base directory for cache storage.
Default base directory for incremental update output.
## Entity types
Default entity types to extract during graph construction.
## Configuration class defaults
The following sections document default values for each configuration class.
### BasicSearchDefaults
Basic search prompt template.
Number of results to return.
Maximum context tokens.
Completion model ID.
Embedding model ID.
### ChunkingDefaults
Chunking strategy type (from ChunkerType enum).
Chunk size in tokens.
Overlap between chunks in tokens.
Encoding model for tokenization.
Metadata to prepend to chunks.
### ClusterGraphDefaults
Maximum size of clusters.
Whether to use the largest connected component.
Random seed for clustering (3735928559 in decimal).
### CommunityReportDefaults
Prompt for graph-based community reports.
Prompt for text-based community reports.
Maximum report length in tokens.
Maximum input length in tokens.
Completion model ID.
Model instance name for caching.
### DriftSearchDefaults
DRIFT search prompt.
Reduce step prompt.
Maximum data tokens.
Maximum reduce tokens.
Temperature for reduce step.
Maximum completion tokens for reduce.
Concurrency level for DRIFT operations.
Number of followup queries.
Number of primer folds.
Maximum tokens for primer LLM.
Search depth.
Text unit proportion for local search component.
Community proportion for local search component.
Top k entities for local search.
Top k relationships for local search.
Maximum data tokens for local search.
Temperature for local search.
Top p for local search.
Number of completions for local search.
Maximum generation tokens for local search.
Maximum completion tokens for local search.
Completion model ID.
Embedding model ID.
### EmbedTextDefaults
Embedding model ID.
Model instance name for caching.
Batch size for embedding operations.
Maximum tokens per batch.
List of embeddings to generate (uses default\_embeddings).
### ExtractClaimsDefaults
Whether claim extraction is enabled.
Claim extraction prompt.
Description of claims to extract.
Maximum number of gleaning iterations.
Completion model ID.
Model instance name for caching.
### ExtractGraphDefaults
Graph extraction prompt.
Entity types to extract.
Maximum number of gleaning iterations.
Completion model ID.
Model instance name for caching.
### TextAnalyzerDefaults
Used for NLP-based graph extraction.
Noun phrase extractor type.
SpaCy model name.
Maximum word length to consider.
Delimiter between words.
Whether to include named entities.
List of nouns to exclude (uses EN\_STOP\_WORDS).
Entity tags to exclude.
Part-of-speech tags to exclude.
Tags for noun phrases.
Grammar rules for noun phrase combination.
Default:
```python theme={null}
{
"PROPN,PROPN": "PROPN",
"NOUN,NOUN": "NOUNS",
"NOUNS,NOUN": "NOUNS",
"ADJ,ADJ": "ADJ",
"ADJ,NOUN": "NOUNS"
}
```
### ExtractGraphNLPDefaults
Whether to normalize edge weights.
Text analyzer configuration.
Number of concurrent requests.
Async mode to use.
### GlobalSearchDefaults
Map step prompt.
Reduce step prompt.
Knowledge generation prompt.
Maximum context tokens.
Maximum data tokens.
Maximum map response length in words.
Maximum reduce response length in words.
Community rating threshold for inclusion.
Keep parent community if children are relevant.
Number of times to rate each community.
Use community summary instead of full context.
Maximum community hierarchy level.
Completion model ID.
### StorageDefaults
Storage type (from StorageType enum).
Text encoding.
Base directory for file storage.
Azure connection string.
Azure container name.
Azure account URL.
Azure CosmosDB account URL.
### InputDefaults
Input type.
Text encoding.
File pattern for matching input files.
Column name for document IDs.
Column name for document titles.
Column name for document text.
### InputStorageDefaults
Extends StorageDefaults.
Base directory for input storage.
### CacheStorageDefaults
Extends StorageDefaults.
Base directory for cache storage.
### CacheDefaults
Cache type.
Cache storage configuration.
### LocalSearchDefaults
Local search prompt.
Text unit proportion.
Community proportion.
Maximum conversation history turns.
Top k entities to retrieve.
Top k relationships to retrieve.
Maximum context tokens.
Completion model ID.
Embedding model ID.
### OutputStorageDefaults
Extends StorageDefaults.
Base directory for output storage.
### PruneGraphDefaults
Minimum node frequency.
Maximum node frequency standard deviation.
Minimum node degree.
Maximum node degree standard deviation.
Minimum edge weight percentage.
Whether to remove ego nodes.
Keep only largest connected component.
### ReportingDefaults
Reporting type.
Base directory for reporting.
Connection string for blob reporting.
Container name for blob reporting.
Storage account blob URL.
### SnapshotsDefaults
Whether to save embedding snapshots.
Whether to save GraphML snapshots.
Whether to save raw graph snapshots.
### SummarizeDescriptionsDefaults
Summarization prompt.
Maximum summary length in tokens.
Maximum input tokens.
Completion model ID.
Model instance name for caching.
### UpdateOutputStorageDefaults
Extends StorageDefaults.
Base directory for update output storage.
### VectorStoreDefaults
Vector store type (from VectorStoreType enum).
Database URI for vector store.
### GraphRagConfigDefaults
Root configuration defaults.
Legacy model configurations.
Completion model configurations.
Embedding model configurations.
Default concurrent requests.
Default async mode.
Reporting configuration defaults.
Input storage configuration defaults.
Output storage configuration defaults.
Update output storage configuration defaults.
Cache configuration defaults.
Input configuration defaults.
Text embedding configuration defaults.
Chunking configuration defaults.
Snapshots configuration defaults.
Graph extraction configuration defaults.
NLP graph extraction configuration defaults.
Description summarization configuration defaults.
Community reports configuration defaults.
Claims extraction configuration defaults.
Graph pruning configuration defaults.
Graph clustering configuration defaults.
Local search configuration defaults.
Global search configuration defaults.
DRIFT search configuration defaults.
Basic search configuration defaults.
Vector store configuration defaults.
Workflows list.