Announcing Native BM25 Ranking in AlloyDB and Cloud SQL
cloud.google.com·1d ago
AI Research
arXiv CS.AI·Jun 9·5 min read
MM-Matryoshka: Towards Budget-Elastic Visual Document Retrieval via a 2D Multimodal Matryoshka Training Framework
TL;DR
Existing visual document retrieval systems struggle with high storage and computational costs due to their reliance on multiple vectors from Vision-Language Models (VLMs). MM-Matryoshka is a new 2D Matryoshka training framework that allows for budget-elastic retrieval by optimizing both vector width and encoder depth.
✦ Why It Matters
Engineers can implement MM-Matryoshka to optimize visual document retrieval systems for better resource management.
Key Takeaways
- Implement MM-Matryoshka for flexible visual document retrieval budgets.
- Reduce storage and computational costs without sacrificing accuracy.
- Utilize a single model for varying retrieval requirements.
- Explore the framework's adaptability across different VDR applications.
How It Works
MM-Matryoshka employs a 2D training framework that allows for simultaneous adjustments in vector width and encoder depth. This dual flexibility enables the model to select optimal retrieval budgets at inference time, enhancing efficiency without the need for multiple models.