gpt-spa (created by Greg DeCarlo - @mrinreality1 on X)
Customizable GPTs stored on your local browser, using OpenAI models text-embedding-ada-002, tts-1, whisper-1, dall-e-3, and gpt-4-vision-preview. Enhanced with Azure OpenAI support!
Create Custom GPTs!
Key Features
- Configure GPTs by specifying system prompts and selecting from files, tools, and other GPT models.
- Utilize local vector database for document retrieval (RAG) without relying on the OpenAI Assistants API.
- Upload image files for analysis using the GPT-4 Vision model. Activate 'Image Generation (DALL-E)' in the Configure tab to generate images based on conversation context.
- Activate 'Text-to-Speech (TTS)' to hear outputs using OpenAI's Text-to-Speech model.
- Share your GPT configurations and conversations easily with a URL, ensuring privacy as data is stored only in browser local storage or the URL hash (no server-side data storage).
Built-in Tools
- Image Generation (DALL-E)
- Text-to-Speech (TTS)
- Bing Search (requires API Key)
- Daily Quotation Generator (powered by GPT)
Usage Instructions
- Enter your OPENAI_API_KEY (stored locally)
- Optionally enter your OPENAI_API_ORG
- Use the sidebar to select or create custom GPTs
- Engage with your GPT using the provided controls. Restart or switch GPTs to initiate new conversations
Advanced Features
- Dynamic conversations with custom GPTs; shareable links for both conversations and GPT configurations
- Upload or drag-and-drop PDFs or text files for enhanced retrieval-based searching
- Upload images and engage with your GPT about their content (local usage only, data is not retained after browser refresh)
- Request DALL-E to create images based on conversation history, including uploaded images
- Use TTS with a choice of voices for an auditory experience
- Switch between multiple GPTs or blend their capabilities
- Intuitive GPT builder interface
- [New] Direct voice inputs to your GPT via a microphone button
- [New] Enhance context through file saving, leveraging vector search for text-based files
- [New] Integrate Bing search results into retrieval processes (RAG)
- [New] Seamlessly transition between models or modalities using tools and GPT combinations (configurable in the Configure tab)
Upcoming Enhancements
- Integration of Whisper Speech-to-Text for conversational inputs
- Enabling GPTs to pose multiple-choice questions and suggest answers
- Integration with Microsoft Graph API for email management, appointment scheduling, and contact searches
- Memory storage and reminder setting functionalities
- Tools for creating charts, diagrams, and tabular data
- Support for connecting to external data sources compatible with CORS
About this repository
- Contributions welcome!
- MIT License
- Author: Greg DeCarlo
- README was modified by ChatGPT