https://github.com/tortillazhawaii/ekspercik
PDF ChatBot based on Large Language Models
https://github.com/tortillazhawaii/ekspercik
chroma gpt-4 langchain llm ollama openai python rag streamlit vector-database
Last synced: 6 months ago
JSON representation
PDF ChatBot based on Large Language Models
- Host: GitHub
- URL: https://github.com/tortillazhawaii/ekspercik
- Owner: TortillaZHawaii
- Created: 2023-12-07T11:14:37.000Z (almost 2 years ago)
- Default Branch: main
- Last Pushed: 2023-12-08T10:41:59.000Z (almost 2 years ago)
- Last Synced: 2023-12-08T17:49:41.996Z (almost 2 years ago)
- Topics: chroma, gpt-4, langchain, llm, ollama, openai, python, rag, streamlit, vector-database
- Language: Python
- Homepage:
- Size: 320 KB
- Stars: 0
- Watchers: 2
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# Ekspercik
Ekspercik is a chat bot similar to ChatGPT that can answer based on provided PDFs.
It uses tesseract OCR to extract text from PDFs (also from images in PDFs) and then stores them persistently to Chroma.
Chat bot then uses this data to answer questions.
Since this was hacked in a day it has few rough edges.
It supports running against OpenAI LLMs such as GPT-4 or GPT-3.5-Turbo, as well as self hosted Ollama models.
I tested it with `gpt-4` and `mistral:7b` models. Models can be changed in a sidebar.
## Usage
Run it with streamlit:
```bash
streamlit run ekspercik.py
```
For now you will need to guess and install all the dependencies yourself 😕.
## Screenshots
