MiG-NJU/EvoEmbedding-4B

EvoEmbedding: Evolvable Embedding for Long-Context Retrieval

🔗 GitHub Repository | 📚 Training Dataset | 📑 Paper (Coming Soon)

EvoEmbedding is a novel embedding model designed for long-context and dynamic retrieval scenarios. Unlike static embedding models that chunk text in isolation, EvoEmbedding maintains a continuously updated Latent Memory Queue. This allows it to capture temporal dynamics and generate context-aware, evolvable embeddings for precise retrieval in agentic workflows and long-conversations.

📦 Model Family

We provide EvoEmbedding in three sizes based on the Qwen architecture:

Model	Parameters	Base Model	Hugging Face Link
EvoEmbedding-0.8B	0.8B	Qwen3.5-0.8B	ClareNie/EvoEmbedding-0.8B
EvoEmbedding-2B	2B	Qwen3.5-2B	ClareNie/EvoEmbedding-2B
EvoEmbedding-4B	4B	Qwen3-4B	ClareNie/EvoEmbedding-4B

📚 Citation

If you find this model or our methodology useful, please cite our paper:

@article{nie2026evoembedding,
  title={Evolvable Embedding for Long-Context Retrieval},
  author={Nie, Chang and Fu, Chaoyou and Shan, Caifeng},
  journal={arXiv preprint},
  year={2026}
}