https://github.com/asaniczka/SERP-RAG-Chatlorin

Starter project for creating an API endpoint that performs web searches, scrapes results, and generates answers using LLM.
https://github.com/asaniczka/SERP-RAG-Chatlorin

chromadb gemini-pro llm rag serp

Last synced: 7 days ago
JSON representation

Starter project for creating an API endpoint that performs web searches, scrapes results, and generates answers using LLM.

• Host: GitHub
• URL: https://github.com/asaniczka/SERP-RAG-Chatlorin
• Owner: asaniczka
• Created: 2024-03-16T09:24:57.000Z (8 months ago)
• Default Branch: main
• Last Pushed: 2024-07-07T12:33:07.000Z (4 months ago)
• Last Synced: 2024-08-02T15:26:05.561Z (3 months ago)
• Topics: chromadb, gemini-pro, llm, rag, serp
• Language: Python
• Homepage:
• Size: 328 KB
• Stars: 2
• Watchers: 2
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# LLM Chat Endpoint with Internet Access

This Python project focuses on creating an API endpoint that:

1. Searches Google for information

2. Scrapes the results

3. Processes the pages for RAG

4. Uses an LLM to generate an answer

5. and returns the response.

## Notes:

### Messages

- Most of the core function have been written to be LLM independent.

- The core uses `BaseMessageLog` and `BaseMessage` to pass around messages and roles.

- Only the Gemini module uses Gemini-specific formats.

- You can replace the Gemini-specific formats with OpenAI chat format and the core should still work.

### Working Mode

- By default, the Dockerfile sets the app to be in low-memory mode.

- This mode will not use JavaScript rendering to load pages.

- The performance is decent, but if you wish to increase the number of pages scraped, set the `WORKING_MODE` env to anything other than `low-mem`.

## How to Use

1. Clone the project repository to your local machine

2. Run `docker build -t serp_rag:latest .`

3. Launch a container with `docker run -p 8080:8000 --name serp_rag_con -e GOOGLE_GEMINI_KEY="YOUR GEMINI KEY" serp_rag:latest`

4. Load up `https://localhost:8080/docs` and try the API.