Skip to content

jyimu/AeloRu

Repository files navigation

AeloRu

Adaptive Elastic Learning with Orthogonal Robust Units

Beyond Low-Rank: Dynamic Plasticity through Amplitude-Direction Decoupling


What is AeloRu?

AeloRu is a research framework for investigating semantic structures in PEFT weight matrices and developing next-generation adaptive low-rank update methods. It combines HiRA, DoRA, Hebbian learning, and asynchronous architectures to enable real-time model alignment with dynamic plasticity.

Core Research Question: How can integrating Hebbian Learning principles with LoRA-based parameter-efficient fine-tuning mitigate catastrophic forgetting during long-term sequential task learning, while enhancing adaptation speed in short-term scenarios?


Research Objectives(the latest progress)

Phase Objective Status
P0 Hidden state semantic analysis ✅ Done(Link)
P1 HiRA-DoRA fusion implementation ✅ Done(Link)
P2 Hebbian-RL hybrid learning ✅ Done(Link)
P3 Training on real data & writing paper 🔄 Doing
P4 Asynchronous PEFT architecture 📋 Planned

Expected Contributions

  1. First HiRA-DoRA fusion with amplitude-direction decoupling on modulation terms
  2. Hebbian-RL hybrid learning for real-time adaptive plasticity
  3. Adapter-level asynchrony enabling zero-latency serving during continuous learning

License

MIT License - see LICENSE for details.

About

AeloRu (Adaptive Elastic Learning with Orthogonal Robust Units) enables real-time, continuous learning on resource-constrained devices. Drop-in memory module for LLMs.

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors