Skip to content

Latest commit

 

History

History
24 lines (19 loc) · 840 Bytes

README.md

File metadata and controls

24 lines (19 loc) · 840 Bytes

code-help

Setup

Ensure that the local Python version is >= 3.8. Install the required libraries:

pip install -r requirements.txt

How to Run

python code_help.py

Reference

Code Referenced from https://github.com/Chitti-Ankith/Story-QA-using-GPT4All/tree/main

What Happens

  • Inputs a file specified by the user (or a default file) and loads it into EvaDB as a string.
  • Then, it creates code embeddings using a transformers model (all-MiniLM-L6-v2).
  • Creates an index using Qdrant to help with searching.
  • Reads question from user.
  • Runs a similarity search on the question throught out the embeddings table and returns a list of the top 5 most similar entries.
  • Uses a GPT4All model (ggml-model-gpt4all-falcon-q4_0.bin) to generate an answer to the user's question based upon those 5 entries.