Add model card for MIA
#1
by nielsr HF Staff - opened
README.md
ADDED
|
@@ -0,0 +1,33 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: mit
|
| 3 |
+
library_name: transformers
|
| 4 |
+
pipeline_tag: text-generation
|
| 5 |
+
---
|
| 6 |
+
|
| 7 |
+
# Memory Intelligence Agent (MIA)
|
| 8 |
+
|
| 9 |
+
Memory Intelligence Agent (MIA) is a memory framework designed for deep research agents (DRAs). It transforms agents from "passive record-keepers" into "active strategists" using a sophisticated **Manager-Planner-Executor** architecture.
|
| 10 |
+
|
| 11 |
+
- **Paper:** [Memory Intelligence Agent](https://huggingface.co/papers/2604.04503)
|
| 12 |
+
- **Repository:** [https://github.com/ECNU-SII/MIA](https://github.com/ECNU-SII/MIA)
|
| 13 |
+
|
| 14 |
+
## Overview
|
| 15 |
+
|
| 16 |
+
MIA replaces traditional "memory dumps" with a specialized architecture to enable efficient reasoning and autonomous evolution:
|
| 17 |
+
|
| 18 |
+
- **The Manager**: A non-parametric memory system that stores and optimizes compressed historical search trajectories to eliminate bloat.
|
| 19 |
+
- **The Planner**: A parametric memory agent that produces search plans and evolves its strategy via Continual Test-Time Learning during inference.
|
| 20 |
+
- **The Executor**: A precision instrument that searches and analyzes information guided by the search plan.
|
| 21 |
+
|
| 22 |
+
MIA employs an alternating reinforcement learning paradigm to enhance cooperation between components and establishes a bidirectional conversion loop between parametric and non-parametric memories.
|
| 23 |
+
|
| 24 |
+
## Citation
|
| 25 |
+
|
| 26 |
+
```bibtex
|
| 27 |
+
@article{qiao2026mia,
|
| 28 |
+
title={Memory Intelligence Agent},
|
| 29 |
+
author={Jingyang Qiao and Weicheng Meng and Yu Cheng and Zhihang Lin and Zhizhong Zhang and Xin Tan and Jingyu Gong and Kun Shao and Yuan Xie},
|
| 30 |
+
journal={arXiv preprint arXiv:2604.04503},
|
| 31 |
+
year={2026}
|
| 32 |
+
}
|
| 33 |
+
```
|