config/config.yaml

# API Keys
google_api_key: "AIzaSyCudDlnOaHwbfOo-ofigIwSn5NSQjRkM6o"
google_cx: "c4935dcef4e4a42cf"
bing_api_key: "your_bing_api_key"
openai_api_key: "sk-BDvOAVHD0svQJyNw1X1dxKRRWihB992xHA5McFCaFfAwbu6t"
deepseek_api_key: "sk-22e0ee691f474267b67eae0d48e962b8"
bocha_api_key: "sk-3108ee39bc1240d899c4bd5f2abef5e7"
bocha_base_url: "https://api.bochaai.com"

# LLM Settings
llm_settings:
  providers:
    - name: "deepseek"
      enabled: true
      model: "deepseek-chat"
      base_url: "https://api.deepseek.com"
      api_key: ${deepseek_api_key}
      models: ["deepseek-chat"]
    - name: "openai"
      enabled: true
      model: "gpt-4o"
      base_url: "https://api.openai.com/v1"  # 使用标准 OpenAI API URL
      api_key: ${openai_api_key}
      models: ["gpt-4o"]
  default_provider: "deepseek"  # 默认使用的提供商

# 检索设置
retrieval_settings:
  default_method: "standard"
  methods:
    - name: "standard"
      enabled: true
      model_settings:
        embedding_model: "BAAI/bge-m3"  # 使用 Hugging Face 模型 ID
        reranker_model: "BAAI/bge-reranker-large"  # 使用 Hugging Face 模型 ID

# Search Settings
max_results: 20
language: "zh-CN"
search_provider: "bocha"

# Document Processing
max_passage_length: 500
min_passage_length: 100

# Vector Search Settings
embedding_model: "BAAI/bge-m3"  # 使用 Hugging Face 模型 ID
reranker_model: "BAAI/bge-reranker-large"  # 使用 Hugging Face 模型 ID
batch_size: 32
use_gpu: true

# Ranking Settings
initial_top_k: 100
final_top_k: 3
retrieval_weight: 0.3
rerank_weight: 0.7

# Search Settings
search_settings:
  trusted_domains:
    - 'discoverhongkong.com'
    - 'tourism.gov.hk'
    - 'hong-kong-travel.com'
    - 'timeout.com.hk'
    - 'openrice.com'
    - 'lcsd.gov.hk'
    - 'hkpl.gov.hk'
  
proxy:
  enabled: false
  host: "127.0.0.1"
  port: 8880