# Configuración de Google Cloud Platform
project_id: "tu-proyecto-gcp"
location: "us-central1"  # o us-east1, europe-west1, etc.
bucket: "tu-bucket-nombre"

# Configuración del índice vectorial
index:
  name: "mi-indice-rag"
  dimensions: 768  # Para text-embedding-005 usa 768
  machine_type: "e2-standard-2"  # Tipo de máquina para el endpoint
  approximate_neighbors_count: 150
  distance_measure_type: "DOT_PRODUCT_DISTANCE"  # O "COSINE_DISTANCE", "EUCLIDEAN_DISTANCE"

# Configuración de embeddings
embedder:
  model_name: "text-embedding-005"
  task: "RETRIEVAL_DOCUMENT"  # O "RETRIEVAL_QUERY" para queries

# Configuración de LLM para chunking
llm:
  model: "gemini-2.0-flash"  # O "gemini-1.5-pro", "gemini-1.5-flash"

# Configuración de chunking
chunking:
  strategy: "contextual"  # "recursive", "contextual", "llm"
  max_chunk_size: 800
  chunk_overlap: 200  # Solo para LLMChunker
  merge_related: true  # Solo para LLMChunker
  extract_images: true  # Solo para LLMChunker