Releases: Zaki-1052/GPTPortal
GPTPortal v1.1.2 Stable Model Update
GPTPortal v1.1.2
Mistral & Google Release
Description
This is the latest stable release for the newest models, including:
- Mistral Tiny with 7 Billion Parameters (Cheapest)
- Mixtral 8x7b with a Mixture of Experts (MoE) Architecture
- Mistral Large, the newest and most intelligent Mistral model
- Release Article from Mistral
- Microsoft Azure Partnership Announcement
- Gemini 1.5 Pro: Early Access Google Model with 1 Million Tokens!
- DeepMind Release Article
- Gemini 1.5 Technical Report
- [Unreleased] Gemini Ultra: Largest model, but generally unavailable
- Gemini Ultra Release Notes & Technical Report
Feel free to raise an issue if you have any more requests for new models!
Update Note: I'm also getting to Embeddings very soon, and will hopefully release shortly.
Be aware that Gemini-1.5-Pro is in Private Preview, so it will not currently function in this portal via API until it is more generally released, likely in March.
Full Changelog: v1.1.1...v1.1.2
GPTPortal v1.1.1 Mistral Stable Release
GPTPortal v1.1.1 Mistral Stable Release
GPTPortal v1.1.1
Mistral Release
Description
This release of my GPTPortal includes an optional Docker Image and Package, along with the addition of three MIstral AI Models: Tiny, Small, and Medium. The Assistants API Branch is complete, and I will create a release for it shortly. I still intend to play around with embeddings on Main, but more likely, they will remain relatively simple, and any file loading will go through the customizable assistants. Please direct any queries of future plans to the discussion of that branch. Thank you!
Release: https://github.com/Zaki-1052/GPTPortal/releases/tag/v1.0-assistants
- Note: I've also upgraded the
node_modulesand fixed an issue with a misdirectedgpt-3.5-turboalias. Be sure to runnpm installif you encounter any network request issues. If you'd like to utilize the OpenAI SDK, direct your attention to the Assistants Branch, or use Google's Gemini Package. - Edit: See this issue for more details on hotfixes: #12 (comment)
- Note: The previous release was deleted for an issue with the differing
user_inputschema for GPT-4-Vision and the Mistral models. Please pull your local version if you were experiencing any problems or configuration errors.
Pull Requests
- Build a multi-platform docker image on push and new releases by @FarisZR in #6
- Bump follow-redirects from 1.15.3 to 1.15.5 by @dependabot in #8
Full Changelog: v1.0.1...v1.1.1
GPT-Voice v1.1.0 Notes
GPT-Voice Portal
Version 1.1.0 Notes
This release of the Voice Branch is simply a merged update with Main that you can easily download. Please direct your attention to the Discussion of v1.1.0 for more details, as described here: https://github.com/Zaki-1052/GPTPortal/discussions/9
What's Changed
- Build a multi-platform docker image on push and new releases by @FarisZR in #6
- Bump follow-redirects from 1.15.3 to 1.15.5 by @dependabot in #8
New Contributors
- @FarisZR made their first contribution in #6
- @dependabot made their first contribution in #8
Full Changelog: v1.0.1...v1.1.0-voice
Contributors
Assets 2
GPTPortal for Assistants API - v1.0
GPTPortal for Assistants API
Version 1.0 - Stable Release
Description
This version of GPTPortal is dedicated solely to the Assistants API, which uses OpenAI's Beta system for automatically managing conversation history and allowing GPT-4 to execute Python in a native Jupyter Notebook. This also allows for reading files of all types. A more thorough description of its capabilities can be found in the ReadMe of this branch.
Please direct any inquiries you may have about its functionality, features, and codebase to this discussion.
Lastly, I intend to see if the native embedding models could serve to be more useful than Assistants, but if this is otherwise not the case, then this branch will be merged as a separate endpoint with main, and users will be able to choose to use an "Assistants Mode" for more exact custom instructions and file uploads, along with Python execution.
Thank you!
What's Changed
- Build a multi-platform docker image on push and new releases by @FarisZR in #6
- Bump follow-redirects from 1.15.3 to 1.15.5 by @dependabot in #8
New Contributors
- @FarisZR made their first contribution in #6
- @dependabot made their first contribution in #8
Full Changelog: v1.0.1...v1.0-assistants
GPTPortal v1.0.1 Latest Stable Release
GPTPortal v1.0.1
Latest Stable Release
Overview
This release of the GPT & Gemini Chatbot Portal is the latest stable version before my implementation of simple Embeddings and RAG, along with the eventual inclusion of Gemini Ultra when Google releases its API in February.
I recommend directing your attention to the Discussion of the Version 1 Release: GPTPortal for any and all questions or concerns about the functionality of this repository.
Description
This release at #4 describes the key features and updates of the project; the attached version simply includes a few corrections to the ReadMe, and an update to the Model Aliases, namely:
- GPT-4-Turbo Preview:
gpt-4-0125-preview- The latest GPT-4 model intended to reduce cases of “laziness” where the model doesn’t complete a task. Learn more.
- GPT-3.5-Turbo:
gpt-3.5-turbo-1025- Even Cheaper Pricing and Bug Fixes
Future Plans
I intend to separately tag my branches of the Voice Only and Assistants API modes, the former of which is complete and the latter is in production. My Vector Embeddings and personal implementation of Retrieval Augmented Generation will first be geared towards a BioMedical application in a branch, and then merged with Main for proper customization.
Due to OpenAI's latest API model upgrades, users will be given the option to either stay with Google Gemini or choose the different Embedding Models with various pricing and functionalities. A compiled version, compatible with Windows, MacOS, and Linux will be available in April, once OpenAI confirms their release of either GPT-4.5 or possibly GPT-5.
Thank you for your patience!
Full Changelog: v1.0.0...v1.0.1
GPT & Gemini ChatBot Portal – v1 Release
Release Notes for "GPT & Gemini ChatBot Portal – v1.0.0"
Overview
The "GPT & Gemini Chat-Bot Portal" is a feature-rich, lightweight Node.js-based web application, designed to interact with the OpenAI GPT-4 API, including its Vision, Hearing, and Speaking capabilities along with image-generation. This version marks the integration of Google’s Gemini models, with coming support for RAG and vector embeddings.
Key Features
- Multi-Modal Interaction: Engage with the chatbot using text, voice, & image inputs.
- Advanced AI Integration: Powered by OpenAI's GPT-4 & Google Gemini models.
- User-Friendly Interface: Simplified and intuitive for seamless user experience.
- Customizable Parameters: Tailor chat responses with system-defined instructions and model settings.
- Diverse Model Selection: Choose from various models including GPT-4, GPT-4-Vision, Gemini-Pro, and more.
- Secure Authentication: Basic authentication ensures secure access.
- Export Functionality: Easily export conversation history in HTML format.
Enhancements & Updates
- Google Gemini Integration: Latest Google Gemini models are now accessible.
- Model Selector Added: Easily switch between different AI models.
- Improved Image Handling: Enhanced image upload and processing capabilities.
- Markdown Rendering: Chat messages are now beautifully rendered in Markdown.
- Voice and Image Integration: Engage in voice conversations and upload images for contextual chats.
- DALL·E 3 Image Generation: Generate images using the "Generate:" command.
Resources
- YouTube Demo: Check out this YouTube demo for a full walkthrough of features and installation.
- Full Changelog: Visit the Changelog for detailed version history.
- ReadMe: The ReadMe provides comprehensive project details along with instructions to use and install the repo.
Community & Support
- Join Our Discord: For support and community interaction, join my Discord server.
- Contribute: Contributions are highly encouraged. Please submit pull requests or issues on GitHub.
License
This project is released under the MIT License, promoting open-source collaboration and innovation.
Stay updated and connected for the most immersive chatbot experience to date!