Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/kevinknights29/local_chatgpt_streamlit
This project aims to create a Local ChatGPT with Streamlit
https://github.com/kevinknights29/local_chatgpt_streamlit
openai-api python streamlit
Last synced: 16 days ago
JSON representation
This project aims to create a Local ChatGPT with Streamlit
- Host: GitHub
- URL: https://github.com/kevinknights29/local_chatgpt_streamlit
- Owner: kevinknights29
- License: mit
- Created: 2023-10-21T15:35:37.000Z (over 1 year ago)
- Default Branch: main
- Last Pushed: 2024-02-29T04:43:31.000Z (12 months ago)
- Last Synced: 2024-11-29T09:29:48.953Z (3 months ago)
- Topics: openai-api, python, streamlit
- Language: Python
- Homepage:
- Size: 26.4 KB
- Stars: 1
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# Local_ChatGPT_Streamlit
This project aims to create a Local ChatGPT with Streamlit
**Topics**
1. [Overview](#overview)
2. [Goals](#goals)
3. [Scope and Context](#scope-and-context)
4. [System Design](#system-design)
5. [Alternatives Considered](#alternatives-considered)
6. [Learning Logs](#learning-logs)
7. [Resources](#resources)
---
## Overview
### Problem Statement
I'm a ChatGPT Plus subscriber, but I've realized that in the past few months I haven't been using ChatGPT enough to justify the 20 USD/month subscription.
OpenAI offers an API to support the `Pay As You Go` model, which I believe is the best option for me.
### OpenAI API Pricing
| Model | Context Length | Price: Input / 1K tokens | Price: Output / 1K tokens |
|--------|----------------|--------------------------|---------------------------|
| gpt3.5 | 4K tokens | 0.0015 USD | 0.002 USD |
| gpt3.5 | 16K tokens | 0.003 USD | 0.004 USD |
| gpt4 | 8K tokens | 0.03 USD | 0.06 USD |
| gpt4 | 32K tokens | 0.06 USD | 0.12 USD |
For more details, please check: [OpenAI Pricing](https://openai.com/pricing)
### Estimations
If I use the API for gpt3.5 the most basic API call should cost 0.0035 (input + output) which would require
~ 5,700 calls to consume 20.00 USD.
We can consider each API call as an interaction with ChatGPT, and I can assure you that I'm not interacting with ChatGPT 5,7k times a month.
## Goals
- Make a cost-effective ChatGPT.
## Scope and Context
The scope is not to recreate all the features in ChatGPT.
The features needed are:
- Chat UI,
- Users should be able to send and receive messages.
- Response Streaming,
- Responses should be displayed as they (tokens) are being generated.
- API should be requested as login (for security purposes).
## System Design
Leverage Streamlit and OpenAI libraries to implement the features needed.
```mermaid
flowchart LR
step_1["Request user's OpenAI API key"]
step_2{"Is OpenAI API key valid?"}
step_3_a["Render Chat UI"]
step_3_b["Display error message"]
step_4["User submits question (prompt) in chat"]
step_5["OpenAI API response is streamed to chat"]
step_1 --> step_2
step_2 -- Yes --> step_3_a
step_2 -- No --> step_3_b
step_3_a --> step_4
step_4 --> step_5
```
## Alternatives Considered
We could build this application using Gradio, but Streamlit has a better deployment process.
For more, reference: [Streamlit Community Cloud](https://streamlit.io/cloud)
## Learning Logs
| Date | Learning |
|------------|-----------------------------------------------------------------------------------------------|
| 10/21/2023 | You need to be careful with assigning `open.api_key` multiple times or at different contexts. |
## Resources
### Articles
- [OpenAI’s 'Pay As You Go' Is the Best Way to Use ChatGPT](https://lifehacker.com/openai-s-pay-as-you-go-is-the-best-way-to-use-chatgpt-1850318349)
### Documentation
- [Build conversational apps](https://docs.streamlit.io/knowledge-base/tutorials/build-conversational-apps)