Cipher AI Vault is an Azle-based proof of concept that seamlessly integrates:
- In-memory VectorDB
- In-memory LLM
- Secure asset storage
- Stable memory data storage
- Cycles-distro top-up
- ic-auth for authentication
This versatile platform showcases the Internet Computer's potential for secure, sandboxed AI development, offering adaptable tools for a wide range of AI-driven applications.
Note: This demo is a proof of concept and not intended for production use. It was developed as part of a Developer Grant from the DFINITY Foundation.
- Demo Canisters and Repositories
- Core Features
- Prerequisites
- Quick Setup Instructions
- Detailed Setup and Deployment
- Roadmap
- License
- Frontend Canister: Main entry point for user interactions
- In-memory VectorDB: Stores and manages embeddings for efficient retrieval
- In-memory LLM: Processes natural language queries and interacts with the VectorDB
- Secure Asset Storage: Dedicated module for secure asset storage
- Secure Data Store: Dedicated canister for storing data in stable memory
- Cycles Distro Canister: Manages cycles and top-ups
- ic-auth: Handles authentication with various wallets (Plug, Stoic, NFID, and Internet Identity)
For the best experience, use a WebGPU enabled browser. We recommend Chrome Canary.
- DFX
- Node.js
- Azle development kit
For setup assistance, refer to:
You will need one of the following wallets:
-
Clone the repository:
git clone https://github.com/supaIC/Cipher-AI-Vault.git cd Cipher-AI-Vault
-
Run the setup script:
npm run setup
Note: You may be prompted to enter your DFX identity password during setup.
For detailed instructions, refer to the following README files within this repository:
- Frontend Canister Setup and Deployment
- Cycles Distro Canister Setup and Deployment
- Data Store Canister Setup and Deployment
- Data Store backup canister
- Edit Data Store file entries
- Multiple in-memory LLM support
- Models stored in asset canisters
- Embeddings backed up in Stable Memory
- Generate a Data File from a document using in-memory LLM
- Generate images to be stored in the Image Store using in-memory Stable Diffusion
This project is licensed under the MIT License - see the LICENSE file for details.