Hy-Memory-30B-A3B

What is Hy-Memory-30B-A3B

Hy-Memory-30B-A3B released by Tencent is a Mixture-of-Experts (MoE) large language model available on Hugging Face.

Key features

• A compact long-term memory writing model (approximately 3B active parameters) for the write path in external memory frameworks
• Directly replaces the generic LLM in Mem0, MemOS and Hy-Memory to decide what information to retain, how to represent it precisely, and how to update the persistent memory store
• Supports Chat, Working and Coding scenarios
• Trained in two stages: Memory On-Policy Distillation and Attributed Memory Reinforcement Learning
• Can be directly migrated to unseen memory frameworks such as EverMemOS

Who it is for

Developers who need to build or optimise memory writing components for external memory systems, as well as AI application developers using the Mem0, MemOS or Hy-Memory frameworks.

How to get started

1. Visit https://huggingface.co/tencent/Hy-Memory-30B-A3B
2. Download model weights and integrate them into the Mem0, MemOS or Hy-Memory frameworks
3. In the external memory framework, replace the original general-purpose LLM with this model as the memory writer

Compiled from public descriptions on the tool's official website / README / launch page. Pricing and user numbers have not been verified; please check the official website. Some sections were drafted with AI assistance. Written in Chinese and English; other languages are machine-translated.