RAG-WIKI

Empowering knowledge exploration and generation through LLamaIndex and RAG on the full WIKI knowledge base.

RAG-wiki.mp4

Startup 🚀

1. Clone this repo git clone https://github.com/lyyf2002/RAG-WIKI

2. Download a subset of WIKI processed by me which only has 200MB text.

3. Ensure those files satisfy the following file hierarchy: (storage is the path that stores the index)

   ROOT
   ├── wiki
   ├── storage
   └── RAG-WIKI
   

4. To process the full WIKI or other data, please follow the 5-7.

5. Download the full WIKI data you like from Wikipedia database backup dump, e.g. https://dumps.wikimedia.org/enwiki/latest/enwiki-latest-pages-articles.xml.bz2 for English.

6. use wikiextractor to get the cleaned text for the wiki dataset.

   wikiextractor -o wiki --json --no-templates enwiki-latest-pages-articles.xml.bz2
   

7. storage will be created by the app.py when you first run it. You can change the path to get different index stored before.

8. cd RAG-WIKI

9. Update the api_base and api_key in app.py. You can get a free key for test at https://github.com/chatanywhere/GPT_API_free

10. install the Dependencies:

    pip install streamlit
    pip install llama-index
    pip install langchain
    

11. run streamlit run app.py