https://github.com/matsjfunke/rag-from-scratch

implemented vector similarity algorithms to understand their inner workings, used local embeddding models
https://github.com/matsjfunke/rag-from-scratch

embeddings large-language-models local-llm ollama python rag vector-similarity

Last synced: 2 months ago
JSON representation

implemented vector similarity algorithms to understand their inner workings, used local embeddding models

• Host: GitHub
• URL: https://github.com/matsjfunke/rag-from-scratch
• Owner: matsjfunke
• Created: 2024-04-18T22:44:48.000Z (about 1 year ago)
• Default Branch: main
• Last Pushed: 2024-04-22T07:20:24.000Z (about 1 year ago)
• Last Synced: 2025-02-02T09:41:31.478Z (4 months ago)
• Topics: embeddings, large-language-models, local-llm, ollama, python, rag, vector-similarity
• Language: Python
• Homepage:
• Size: 11.9 MB
• Stars: 3
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
[More on how I personally think about the algorithms here](ALGORITHMS.md)

# How to run scripts, and experiment

1. **clone repo**

   ```bash

   git clone https://github.com/matsjfunke/rag-from-scratch.git

   ```

2. **install dependences**

   ```bash

   python3 -m venv env

   source env/bin/activate

   pip install -r requirements.txt

   ```

3. **run scripts**

   - will take some time on the first run because embeddings aren’t cached (have a look at the handle_embeddings() function to understand)

   ```bash

   python dot-product-rag.py

   ```

5. **tinkering optiions**

   - each script contains a main() function this function gives you inside on tweaks you can make

   - different data: upload other .txt files in the root dir of the repo and specify thier name in file_name variable of main()

   - different prompt: change prompt vairable in main()