# RADIOv2.5: Improved Baselines for Agglomerative Vision Foundation Models

arXiv: https://arxiv.org/abs/2412.07679 · HTML: https://mlsanity.com/paper/2412.07679/

## Key facts

- Cited by 89 papers in the map (data through 14 Oct 2026).
- Builds on 45 of its 52 references that are in the map.
- First posted on arXiv 10 Dec 2024.

## Builds on (most cited first, 30 of 45)

- [Learning Transferable Visual Models From Natural Language Supervision](https://mlsanity.com/paper/2103.00020/) (2021)
- [An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale](https://mlsanity.com/paper/2010.11929/) (2020)
- [ImageNet Large Scale Visual Recognition Challenge](https://mlsanity.com/paper/1409.0575/) (2014)
- [Segment Anything](https://mlsanity.com/paper/2304.02643/) (2023)
- [Emerging Properties in Self-Supervised Vision Transformers](https://mlsanity.com/paper/2104.14294/) (2021)
- [Layer Normalization](https://mlsanity.com/paper/1607.06450/) (2016)
- [Qwen2-VL: Enhancing Vision-Language Model's Perception of the World at Any Resolution](https://mlsanity.com/paper/2409.12191/) (2024)
- [Gaussian Error Linear Units (GELUs)](https://mlsanity.com/paper/1606.08415/) (2016)
- [Sigmoid Loss for Language Image Pre-Training](https://mlsanity.com/paper/2303.15343/) (2023)
- [Vision Transformers for Dense Prediction](https://mlsanity.com/paper/2103.13413/) (2021)
- [Towards VQA Models That Can Read](https://mlsanity.com/paper/1904.08920/) (2019)
- [ChartQA: A Benchmark for Question Answering about Charts with Visual and Logical Reasoning](https://mlsanity.com/paper/2203.10244/) (2022)
- [Depth Anything: Unleashing the Power of Large-Scale Unlabeled Data](https://mlsanity.com/paper/2401.10891/) (2024)
- [DocVQA: A Dataset for VQA on Document Images](https://mlsanity.com/paper/2007.00398/) (2020)
- [Semantic Understanding of Scenes through the ADE20K Dataset](https://mlsanity.com/paper/1608.05442/) (2016)
- [How Far Are We to GPT-4V? Closing the Gap to Commercial Multimodal Models with Open-Source Suites](https://mlsanity.com/paper/2404.16821/) (2024)
- [Token Merging: Your ViT But Faster](https://mlsanity.com/paper/2210.09461/) (2022)
- [ShareGPT4V: Improving Large Multi-Modal Models with Better Captions](https://mlsanity.com/paper/2311.12793/) (2023)
- [Vision Transformers Need Registers](https://mlsanity.com/paper/2309.16588/) (2023)
- [Cambrian-1: A Fully Open, Vision-Centric Exploration of Multimodal LLMs](https://mlsanity.com/paper/2406.16860/) (2024)
- [VILA: On Pre-training for Visual Language Models](https://mlsanity.com/paper/2312.07533/) (2023)
- [DataComp: In search of the next generation of multimodal datasets](https://mlsanity.com/paper/2304.14108/) (2023)
- [OCRBench: On the Hidden Mystery of OCR in Large Multimodal Models](https://mlsanity.com/paper/2305.07895/) (2023)
- [InfographicVQA](https://mlsanity.com/paper/2104.12756/) (2021)
- [Data Filtering Networks](https://mlsanity.com/paper/2309.17425/) (2023)
- [Patch n' Pack: NaViT, a Vision Transformer for any Aspect Ratio and Resolution](https://mlsanity.com/paper/2307.06304/) (2023)
- [Token Merging for Fast Stable Diffusion](https://mlsanity.com/paper/2303.17604/) (2023)
- [Probing the 3D Awareness of Visual Foundation Models](https://mlsanity.com/paper/2404.08636/) (2024)
- [EfficientViT: Memory Efficient Vision Transformer with Cascaded Group Attention](https://mlsanity.com/paper/2305.07027/) (2023)
- [AM-RADIO: Agglomerative Vision Foundation Model -- Reduce All Domains Into One](https://mlsanity.com/paper/2312.06709/) (2023)

## Built on by (most cited first, 30 of 89)

- [Perception Encoder: The best visual embeddings are not at the output of the network](https://mlsanity.com/paper/2504.13181/) (2025)
- [MapAnything: Universal Feed-Forward Metric 3D Reconstruction](https://mlsanity.com/paper/2509.13414/) (2025)
- [What matters for Representation Alignment: Global Information or Spatial Structure?](https://mlsanity.com/paper/2512.10794/) (2025)
- [Eagle 2: Building Post-Training Data Strategies from Scratch for Frontier Vision-Language Models](https://mlsanity.com/paper/2501.14818/) (2025)
- [DINO-Foresight: Looking into the Future with DINO](https://mlsanity.com/paper/2412.11673/) (2024)
- [Visual Representation Alignment for Multimodal Large Language Models](https://mlsanity.com/paper/2509.07979/) (2025)
- [DUNE: Distilling a Universal Encoder from Heterogeneous 2D and 3D Teachers](https://mlsanity.com/paper/2503.14405/) (2025)
- [Nemotron 3 Nano Omni: Efficient and Open Multimodal Intelligence](https://mlsanity.com/paper/2604.24954/) (2026)
- [LocateAnything: Fast and High-Quality Vision-Language Grounding with Parallel Box Decoding](https://mlsanity.com/paper/2605.27365/) (2026)
- [Franca: Nested Matryoshka Clustering for Scalable Visual Representation Learning](https://mlsanity.com/paper/2507.14137/) (2025)
- [Thinking with Camera: A Unified Multimodal Model for Camera-Centric Understanding and Generation](https://mlsanity.com/paper/2510.08673/) (2025)
- [NVIDIA Nemotron Nano V2 VL](https://mlsanity.com/paper/2511.03929/) (2025)
- [Native-Resolution Image Synthesis](https://mlsanity.com/paper/2506.03131/) (2025)
- [SparseVILA: Decoupling Visual Sparsity for Efficient VLM Inference](https://mlsanity.com/paper/2510.17777/) (2025)
- [From Seeing to Experiencing: Scaling Navigation Foundation Models with Reinforcement Learning](https://mlsanity.com/paper/2507.22028/) (2025)
- [C-RADIOv4 (Tech Report)](https://mlsanity.com/paper/2601.17237/) (2026)
- [INSID3: Training-Free In-Context Segmentation with DINOv3](https://mlsanity.com/paper/2603.28480/) (2026)
- [Representation Distribution Matching for One-Step Visual Generation](https://mlsanity.com/paper/2607.02375/) (2026)
- [UniUGG: Unified 3D Understanding and Generation via Geometric-Semantic Encoding](https://mlsanity.com/paper/2508.11952/) (2025)
- [Chorus: Multi-Teacher Pretraining for Holistic 3D Gaussian Scene Encoding](https://mlsanity.com/paper/2512.17817/) (2025)
- [Efficient Universal Perception Encoder](https://mlsanity.com/paper/2603.22387/) (2026)
- [AnyTask: an Automated Task and Data Generation Framework for Advancing Sim-to-Real Policy Learning](https://mlsanity.com/paper/2512.17853/) (2025)
- [NAF: Zero-Shot Feature Upsampling via Neighborhood Attention Filtering](https://mlsanity.com/paper/2511.18452/) (2025)
- [Pathryoshka: Compressing Pathology Foundation Models via Multi-Teacher Knowledge Distillation with Nested Embeddings](https://mlsanity.com/paper/2511.23204/) (2025)
- [Diffusing in the Right Space: A Systematic Study of Latent Diffusability](https://mlsanity.com/paper/2606.03578/) (2026)
- [RADSeg: Unleashing Parameter and Compute Efficient Zero-Shot Open-Vocabulary Segmentation Using Agglomerative Models](https://mlsanity.com/paper/2511.19704/) (2025)
- [OpenVision 3: A Family of Unified Visual Encoder for Both Understanding and Generation](https://mlsanity.com/paper/2601.15369/) (2026)
- [What Matters for Scalable and Robust Learning in End-to-End Driving Planners?](https://mlsanity.com/paper/2603.15185/) (2026)
- [Schrodinger Audio-Visual Editor: Object-Level Audiovisual Removal](https://mlsanity.com/paper/2512.12875/) (2025)
- [MonkeyOCRv2: A Visual-Text Foundation Model for Document AI](https://mlsanity.com/paper/2607.11562/) (2026)

## Abstract (arXiv)

Agglomerative models have recently emerged as a powerful approach to training vision foundation models, leveraging multi-teacher distillation from existing models such as CLIP, DINO, and SAM. This strategy enables the efficient creation of robust models, combining the strengths of individual teachers while significantly reducing computational and resource demands. In this paper, we thoroughly analyze state-of-the-art agglomerative models, identifying critical challenges including resolution mode shifts, teacher imbalance, idiosyncratic teacher artifacts, and an excessive number of output tokens. To address these issues, we propose several novel solutions: multi-resolution training, mosaic augmentation, and improved balancing of teacher loss functions. Specifically, in the context of Vision Language Models, we introduce a token compression technique to maintain high-resolution information within a fixed token count. We release our top-performing variants at multiple scales (-B, -L, -H, and -g), along with inference code and pretrained weights