File size: 1,242 Bytes
57c12cc
ef1baf9
1a6cfa1
ef1baf9
 
 
0ea78e2
57c12cc
 
 
 
609295f
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
---
title: "RizzConn Answering Machine"
emoji: "🥲"
colorFrom: indigo
colorTo: gray
sdk: gradio
sdk_version: 4.36.1
app_file: app.py
pinned: false
---

I attended THE LLM conference and while this repo is just a way to implement the ideas presented in the conference. 
I made a QA system which takes the text from the transcripts and returns a coherent answer in a markdown format. 

The RAG uses embdedding + full text search + reranking to get relevant results. 

Each query is recorded for analysis purpose in the sqlite db. Need to implement a analysis ui. 

A lot of work needs to be done based on the conference but it's a start. Some screenshots. 

![Screenshot 2024-06-13 at 3 47 27 PM](https://github.com/avs20/Rizzconn-Answering-Machine/assets/9055117/07b1ca32-f004-4ea1-b5d5-d75af701842b)
![Screenshot 2024-06-13 at 3 49 30 PM](https://github.com/avs20/Rizzconn-Answering-Machine/assets/9055117/5b434f75-08ed-41be-a11b-189aaa070b2e)
![Screenshot 2024-06-13 at 3 50 33 PM](https://github.com/avs20/Rizzconn-Answering-Machine/assets/9055117/0d9da8fe-d065-472c-9085-84ba25e751f5)
![Screenshot 2024-06-13 at 3 52 54 PM](https://github.com/avs20/Rizzconn-Answering-Machine/assets/9055117/2ef9ab27-ac6d-4e28-9e73-e75bfbe40ba0)