pipeline_settings.yaml

# Copyright (c) Microsoft Corporation.
# Licensed under the MIT License.

# this yaml file serves as a configuration template for the graphrag indexing jobs
# some values are hardcoded while others will be dynamically set
input:
  type: blob
  file_type: text
  file_pattern: .*\.txt$
  storage_account_blob_url: $STORAGE_ACCOUNT_BLOB_URL
  container_name: PLACEHOLDER
  base_dir: .

storage:
  type: blob
  storage_account_blob_url: $STORAGE_ACCOUNT_BLOB_URL
  container_name: PLACEHOLDER
  base_dir: output

reporting:
  type: blob
  storage_account_blob_url: $STORAGE_ACCOUNT_BLOB_URL
  container_name: PLACEHOLDER
  base_dir: logs

cache:
  type: blob
  storage_account_blob_url: $STORAGE_ACCOUNT_BLOB_URL
  container_name: PLACEHOLDER
  base_dir: cache

llm:
  type: azure_openai_chat
  api_base: $GRAPHRAG_API_BASE
  api_version: $GRAPHRAG_API_VERSION
  model: $GRAPHRAG_LLM_MODEL
  deployment_name: $GRAPHRAG_LLM_DEPLOYMENT_NAME
  cognitive_services_endpoint: $GRAPHRAG_COGNITIVE_SERVICES_ENDPOINT
  model_supports_json: True
  tokens_per_minute: 80000
  requests_per_minute: 480
  thread_count: 50
  concurrent_requests: 25

parallelization:
  stagger: 0.25
  num_threads: 10

async_mode: threaded

embeddings:
  async_mode: threaded
  llm:
    type: azure_openai_embedding
    api_base: $GRAPHRAG_API_BASE
    api_version: $GRAPHRAG_API_VERSION
    batch_size: 16
    model: $GRAPHRAG_EMBEDDING_MODEL
    deployment_name: $GRAPHRAG_EMBEDDING_DEPLOYMENT_NAME
    cognitive_services_endpoint: $GRAPHRAG_COGNITIVE_SERVICES_ENDPOINT
    tokens_per_minute: 350000
    concurrent_requests: 25
    requests_per_minute: 2100
    thread_count: 50
    max_retries: 50
  parallelization:
    stagger: 0.25
    num_threads: 10
  vector_store:
    type: azure_ai_search
    collection_name: PLACEHOLDER
    title_column: name
    overwrite: True
    url: $AI_SEARCH_URL

entity_extraction:
  prompt: PLACEHOLDER
  entity_types: PLACEHOLDER

snapshots:
  graphml: True