Comment on page

Phases and Milestones

An outline of ELNA's development phases and associated milestones
ELNA's development roadmap outlines key phases and milestones as we work toward democratized AI on the Internet Computer. Below is a roadmap of our journey:

Phase 1: Foundation Building

Milestone 1: Wallet Integration
  • Implement wallet integration to facilitate user interactions with AI assistants on the canister.
Milestone 2: Vector Embedding for User Data
  • Develop and deploy vector embedding mechanisms within canisters to optimize data storage and retrieval.
  • Integrate Retrieval Augmented Generation (RAG) techniques to enhance data retrieval capabilities.

Phase 2: Inference Engine Deployment

Milestone 3: Full Inference Engine within Canisters
  • Successfully deploy ELNA's full inference engine within canisters on the Internet Computer.
  • Users experience reduced latency and improved AI responsiveness.

Phase 3: Fine-Tuning and Optimization

Milestone 4: Fine-Tuning Capabilities
  • Introduce fine-tuning capabilities within canisters, allowing users to customize AI models.
  • Implement methods like QloRA (Quantization-based Learning of Rate Allocation) for efficient weight storage.

Phase 4: Standalone LLM Agents

Milestone 5: Standalone Large Language Models (LLMs)
  • Explore the development of standalone LLM agents within the Internet Computer ecosystem.
  • Develop advanced AI agents capable of independent operation, akin to state-of-the-art LLMs.
These milestones represent our commitment to pushing the boundaries of AI development within a decentralized and democratized framework. While specific timelines may vary, ELNA is dedicated to achieving these milestones as we work towards our vision.
In subsequent sections, we explore the challenges faced by ELNA and our strategies for community engagement and adoption.