programmingninjas / gpt-3.5-on-steroids

GPT-3.5-ON-STEROIDS combines GPT with Python tools, empowering dynamic web scraping, language processing, and data retrieval. Contribute to advancing text generation with AI. 🚀

GPT 3.5 ON STEROIDS: Autonomous Agent with knowledge beyond 2021

Welcome to GPT 3.5 ON STEROID, an open-source project that enhances the capabilities of GPT by integrating it with various Python libraries and APIs for advanced text generation.

Requirements

Make sure you have the following Python libraries installed:

• openai
• google-serp-api
• tiktoken
• wikipedia
• trafilatura
• streamlit
• google-search-results
• python-dotenv
• youtube-transcript-api
• openpyxl
• PyPDF2
• python-docx
• pandasai

Installation

To install the required packages, run the following command in your terminal:

pip install -r requirements.txt

Additionally, you'll need API keys for the following services:

• SerpAPI
• OpenAI

Running Streamlit

To run the Streamlit application, execute the following command in your terminal:

streamlit run ./app/Chat.py

Integrated Python Functions (Tools)

GPT 3.5 ON STEROID incorporates various Python functions that GPT can call and use, including:

• Web Scraping: Utilizing google-serp-api and trafilatura for dynamic data retrieval.
• Natural Language Processing: Using tiktoken for language processing tasks.
• Information Retrieval: Accessing data from wikipedia for comprehensive information retrieval.
• User Interface: Employing streamlit for creating a user-friendly interface.

Note: Whenever a new tool is added, please ensure the following:

• Update the requirements.txt file to include the new tool/library.
• Update the README.md file to document the newly added tool and its functionality.
• Ensure that your feature does not break the application test before merging.

Contribution Guidelines

We welcome contributions from the community to make GPT 3.5 ON STEROID even better! Please follow these guidelines:

1. Create an Issue: First, create an issue detailing the feature, bug fix, or improvement you plan to work on. Wait for approval and assignment before proceeding to the next step.

2. Assign Yourself: After your issue is approved, get yourself assigned to it. This helps avoid duplication of efforts and ensures everyone is aware of ongoing work.

3. Create a Pull Request (PR): Once assigned, proceed to create your PR. Ensure to mention the assigned issue number in the PR description to link it properly.

Note: PRs without assigned issues will be considered spammy and may lead to disqualification.

4. Fork the repository and create your branch: git checkout -b feature/new-contribution

5. Make your changes and test thoroughly.

6. Commit your changes: git commit -m "Add a brief description of your changes"

7. Push to your forked repository: git push origin feature/new-contribution

8. Create a pull request to the main repository with proof of work attached.

Code of Conduct

Please review our Code of Conduct to understand the community standards.

License

This project is licensed under the MIT License - see the LICENSE.md file for details.

gpt-3.5-on-steroids's Issues

Uploading and Analyzing Files

Use Streamlit file_uploader and analyze file formats such as pdf, word and excel.
You can use python modules like PyPDF2, python-docx and openpyxl and show a visual summary of the data contained.

Add Another Open Source LLM Support

As getting OpenAI key is pain because it expires every 3 months and getting a new one needs a new phone number for verification. So adding option for an open source LLM will be great. Allow user to choose their LLM from a dropdown in Streamlit.

Adding text-to-speech in Streamlit

Adding text-to-speech feature will improve our frontend and make the experience better.

[Feature Request]: add codeql workflow

Is your feature request related to a problem? Please describe.

The repository contains code in python, it does not have workflow for code scanning.

Describe the solution you'd like

I want to add the codeql workflow to automate security checks. CodeQL is the code analysis engine developed by GitHub to identify vulnerabilities in code. It will analyze your code and display the results as code scanning alerts. It will be enabled on every push, commit and pull request using GitHub actions.

Record

• I agree to follow this project's Code of Conduct 
• I'm a SWOC 4.0 contributor
• I want to work on this issue

Tool: OCR using Tesseract

Python OCR is a technology that recognizes and pulls out text in images like scanned documents and photos using Python. It can be completed using the open-source OCR engine Tesseract. By integrating it our agent will be able to analyze and understand images better which will increase it's overall intelligence.
Reference

Creating Data Visualization Tools.

As in our Issue #11 , We added the ability to read various data formats. Now we need to create tools to represent the analysis to user in a visual way by using charts etc. The type of visualization tool can be decided by the agent. Function should be there to plot it.

Better error handling for errors caused due to hallucinations.

Hallucination in AI refers to the generation of outputs that may sound plausible but are either factually incorrect or unrelated to the given context. It may assume file paths and commands that do not exist. So we need better error handling mechanism for this.

Feature: Summarise YouTube Videos

This feature will let users get a summary of most YouTube videos having English subtitles. The program will first fetch the subtitles of the video mentioned by the user, and then, GPT will summarise and show it to the user.

Also provide the final output to the user instead of just thoughts.

Right now, the app just displays the "thoughts" of GPT-3.5. When the user asks questions, it does not give them the final output after completion. I think it might be beneficial for users if the program lets them know that these are just thoughts and not the actual answer their question.

I've thought of two solutions to this problem.

1. We can go the "Bing" way by showing the user text like "Searching on google", "Searching Wikipedia", "Writing to file", "Reading from file" instead of showing the entire thought process of GPT. After it completes, we can show the final output to the user.
2. We can hide the thoughts entirely and just show a loading indicator. After it completes, we can show the final output to the user.

What do you think? :D

Integration of Vector Database.

For managing long term memory of the agent. Integration of a vector database is required.
Currently we write it to a text file which agent can refer to by reading it. Which only helps the agent temporarily during the specific task.
Reference

Feat: Implement automated reviewer assignment workflow for Pull Requests

Describe the feature

This issue addresses the need for a GitHub Actions workflow that automatically assigns reviewers to pull requests upon their opening. Automating the reviewer assignment process will streamline code review processes and ensure timely feedback from team members.

Proposed Feature:

The proposed feature involves the creation of a GitHub Actions workflow that triggers upon the opening of a pull request. This workflow will automatically assign designated reviewers based on predefined list of reviewers in the workflow config.

Expected Behavior:

• When a pull request is opened, the workflow should trigger automatically.
• Based on defined rules or configurations, reviewers will be assigned to the pull request.
• Reviewer assignments should be customizable and adaptable to project-specific requirements.

Benefits:

• Streamlines the code review process by automating reviewer assignments.
• Ensures timely feedback and reduces bottlenecks in the development workflow.
• Improves collaboration and accountability among team members.

Record

• I agree to follow this project's Code of Conduct
• I'm a SWOC24 contributor
• I want to work on this issue

Tool: Execute Python Script

Our agent is capable of reading/writing python scripts but it lacks the capability to execute them on the server. So create a function execute_python_file (filepath,requried_data):

   return output

Plots made with Pandas AI not appearing in Streamlit

In the feature of file analysis. We are using PandasAI for the analysis of Excel files. So plots are not appearing in the streamlit frontend even after using the StreamlitMiddleware of pandasai.