config: config.yaml
njobs: 12

search_service_name: https://search-sales2c-ai-prd.search.azure.cn
search_admin_key: ev6B0OtF66WkDmQKJBa4n1Haa8e8p8N3zdaEBnbWtoAzSeAMWSid


embedding_model_endpoint: http://sales2c-ai.chinanorth3.cloudapp.chinacloudapi.cn/v1-openai/embeddings
embedding_model_key: gpustack_0e3d5b35adaf239b_99adacd6f540c7d81006365c8030b16c
VECTOR_DIMENSION: 4096
FLAG_AOAI: "V3"
FLAG_EMBEDDING_MODEL: qwen3-embedding-8b


extract_method: di+vision-llm
form_rec_resource: https://di-sales2c-ai-prd.cognitiveservices.azure.cn/
form_rec_key: G0vhH3twd5K3YYCgfnttf5V6XTMMU4PMdVvRHsgaTb8kZDoU8ZHjJQQJ99BDAEHpCsCfT1gyAAALACOGmOcn
di-Formulas: false
di-hiRes: true
di_allow_features_ext: pdf;jpep;jpg;png;bmp;tiff;heif


FIGURE_BLOB_ACCOUNT_URL: https://sasales2caiprd.blob.core.chinacloudapi.cn/extracted-image-cat-prd?sp=racwdl&st=2025-08-04T06:34:42Z&se=2035-08-04T14:49:42Z&spr=https&sv=2024-11-04&sr=c&sig=t0DTjfht%2FNaPlXUtxhKr40NzZY5kWovgNxJUeAepvgA%3D


DI_BLOB_ACCOUNT_URL: https://sasales2caiprd.blob.core.chinacloudapi.cn/di-result-cat-prd?sp=racwdl&st=2025-08-04T06:34:11Z&se=2035-08-04T14:49:11Z&spr=https&sv=2024-11-04&sr=c&sig=26wxy5M9lcIO2o9zzr6jOtdw2gQTZnGmampHx5EyXbo%3D


DB_URI: postgresql://pgadmin:vwb54pSQDp8vYkusKms@pg-sales2c-ai-prd.postgres.database.chinacloudapi.cn/document-ai-indexer

# Image understanding
figure_caption:
  include_di_content: false         # Figure content that quotes the result of di
  description_gen_max_images: 0     # The maximum number of images to be described. 0 means no description
  model_endpoint: null
  model_key: null
  model: null                       # azure openai set null
  azure_deployment: gpt-4o          # azure openai deployment name,Other platforms are set to empty
  api_version: 2024-08-01-preview   # azure openai deployment name,Other platforms are set to empty


header_fix: true