retriever:
  retriever_hf_paths:
    RAGatouille: "XThomasBU/Colbert_Index"

metadata:
  metadata_links:
    [
      "https://dl4ds.github.io/sp2025/",
      "https://dl4ds.github.io/sp2025/schedule/",
    ]
  slide_base_link:
    "https://dl4ds.github.io"

    # Assignment base link is used to find the webpage where the assignment is described/posted
  assignment_base_link: "https://tools4ds.github.io/fa2024/assignments/"

  # Define content types - assignments, lectures, etc.
  content_types:
    - "lectures"
    - "assignments"
    - "discussion"
    - "other"

  # These need to be patterns from URLs that can be used to identify the type of content uniquely
  lectures_pattern: "/lectures/"
  assignments_pattern: "/assignments/"
  discussion_pattern: "/discussion/"
  project_pattern: "/project/"

  # These are fields that can be extracted from the webpages of the course content
  lecture_metadata_fields:
    - "title"
    - "tldr"
    - "date"
    - "lecture_recording"
    - "suggested_readings"

  assignment_metadata_fields:
    - "title"
    - "release_date"
    - "due_date"
    - "source_file"

token_config:
  cooldown_time: 60
  regen_time: 180
  tokens_left: 50000
  all_time_tokens_allocated: 1000000

content:
  notebookheaders_to_split_on:
    - ["##", "Section"]
    - ["#", "Title"]

misc:
  github_repo: "https://github.com/edubotics-ai/edubot-core"
  docs_website: "https://dl4ds.github.io/dl4ds_tutor/"

api_config:
  timeout: 60