ZerolanData

Define data classes used in ZerolanLiveRobot and ZerolanCore.

All data classes inherits pydantic.BaseModel with strict schema validation.

See documentation here. The documentation is automatically generated and is consistent with the comments and annotations in the code.

The project structure is as follows:

docs
└─ doc_en.md                # Documentation for zerolan-data.
src
└─ zerolan
   └─ data
      ├─ __init__.py
      ├─ protocol
      │  ├─ protocol.py     # JSON+WebSocket message object
      │  ├─ __init__.py
      ├─ pipeline
      │  ├─ abs_data.py     # Abstract pipeline query and prediction
      │  ├─ asr.py          # Automatic speech recognition
      │  ├─ defense.py      # Defense model
      │  ├─ img_cap.py      # Image capitioning
      │  ├─ llm.py          # Large language model
      │  ├─ milvus.py       # Vector database (Milvus)
      │  ├─ ocr.py          # Optical character recognition
      │  ├─ tts.py          # Text to speech
      │  ├─ vid_cap.py      # Video capitioning
      │  ├─ vla.py          # Vision language action model
      │  └─ __init__.py
      └─ data
         ├─ danmaku.py      # Live-streaming data for Bilibili, YouTube and Twitch
         ├─ prompt.py       # TTS prompt
         ├─ state.py        # Service state
         └─ __init__.py

How to generate documentation?

Run the following commands:

python ./docs/scripts/doc_gen.py