config_dev.yaml

common:
  log_file_path: !ENV ${LOG_FILE_PATH:/data/logs/common.log}
  log_sql_path: !ENV ${SQLALCHEMY_DATABASE_URL:sqlite:////data/logs.db}
  log_level: !ENV ${LOG_LEVEL:INFO}

bd:
  entities:
    # Варианты: fixed_size, sentence, paragraph, blm_sentence, blm_paragraph
    strategy_name: !ENV ${ENTITIES_STRATEGY_NAME:paragraph}
    strategy_params:
    #   words_per_chunk: 50
    #   overlap_words: 25
    #   respect_sentence_boundaries: true
    process_tables: true
    neighbors_max_distance: 1

  search:
    use_qe: true
    use_vector_search: true
    vectorizer_path: !ENV ${EMBEDDING_MODEL_PATH:BAAI/bge-m3}
    device: !ENV ${DEVICE:cuda}
    max_entities_per_message: 150
    max_entities_per_dialogue: 300

  files:
    empty_start: true
    documents_path: /data/documents

llm:
  base_url: !ENV ${LLM_BASE_URL:https://api.deepinfra.com}
  api_key_env: !ENV ${API_KEY_ENV:DEEPINFRA_API_KEY}
  model: !ENV ${MODEL_NAME:meta-llama/Llama-3.3-70B-Instruct}
  tokenizer_name: !ENV ${TOKENIZER_NAME:unsloth/Llama-3.3-70B-Instruct}
  temperature: 0.14
  top_p: 0.95
  min_p: 0.05
  frequency_penalty: -0.001
  presence_penalty: 1.3
  seed: 42