# Eddy-VL 1.9B: Structural Pruning and Layered Distillation for Edge-Deployable Multimodal Embedding

arXiv: https://arxiv.org/abs/2607.16316 · HTML: https://mlsanity.com/paper/2607.16316/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Multimodal distillation, one of 12 directions in Knowledge distillation.
- Builds on 16 of its 20 references that are in the map.
- First posted on arXiv 15 Jul 2026.

## Placed in

- Knowledge distillation / [Multimodal distillation](https://mlsanity.com/distill/multimodal-distillation/)

## Builds on (most cited first, 16 of 16)

- [Microsoft COCO: Common Objects in Context](https://mlsanity.com/paper/1405.0312/) (2014)
- [Distilling the Knowledge in a Neural Network](https://mlsanity.com/paper/1503.02531/) (2015)
- [FlashAttention: Fast and Memory-Efficient Exact Attention with IO-Awareness](https://mlsanity.com/paper/2205.14135/) (2022)
- [Similarity of Neural Network Representations Revisited](https://mlsanity.com/paper/1905.00414/) (2019)
- [Relational Knowledge Distillation](https://mlsanity.com/paper/1904.05068/) (2019)
- [When and why vision-language models behave like bags-of-words, and what to do about it?](https://mlsanity.com/paper/2210.01936/) (2022)
- [Winoground: Probing Vision and Language Models for Visio-Linguistic Compositionality](https://mlsanity.com/paper/2204.03162/) (2022)
- [Matryoshka Representation Learning](https://mlsanity.com/paper/2205.13147/) (2022)
- [SugarCrepe: Fixing Hackable Benchmarks for Vision-Language Compositionality](https://mlsanity.com/paper/2306.14610/) (2023)
- [Qwen3-VL-Embedding and Qwen3-VL-Reranker: A Unified Framework for State-of-the-Art Multimodal Retrieval and Ranking](https://mlsanity.com/paper/2601.04720/) (2026)
- [VLM2Vec: Training Vision-Language Models for Massive Multimodal Embedding Tasks](https://mlsanity.com/paper/2410.05160/) (2024)
- [Evaluation of Deep Convolutional Nets for Document Image Classification and Retrieval](https://mlsanity.com/paper/1502.07058/) (2015)
- [VLM2Vec-V2: Advancing Multimodal Embedding for Videos, Images, and Visual Documents](https://mlsanity.com/paper/2507.04590/) (2025)
- [Rethinking Centered Kernel Alignment in Knowledge Distillation](https://mlsanity.com/paper/2401.11824/) (2024)
- [With Limited Data for Multimodal Alignment, Let the STRUCTURE Guide You](https://mlsanity.com/paper/2506.16895/) (2025)
- [MR2-Bench: Going Beyond Matching to Reasoning in Multimodal Retrieval](https://mlsanity.com/paper/2509.26378/) (2025)

## Abstract (arXiv)

In this report, we introduce Eddy-VL 1.9B, a compressed multimodal embedding model built on Qwen3-VL-Embedding-2B for offline, edge-deployable vision-language retrieval. Eddy-VL targets air-gapped forensic and investigative settings where cloud APIs are unavailable and low latency is essential. Compression combines (i) probe-driven structural pruning that removes four redundant text-decoder layers (28 to 24) ranked by adjacent-layer linear CKA, and (ii) layered knowledge distillation with hole-covering teacher-student mappings, mid-layer attention-map 1-CKA, and final-layer MSE and cosine losses with Matryoshka dimensions {128, 256, 512, 1024, 2048}. The released model contains 1,926,188,032 parameters (3.85 GB bf16), representing approximately 9.5% fewer parameters than the 2.13B teacher model. Empirical evaluations on MMEB-V2 (78 tasks, VLM2Vec protocol) show that Eddy-VL achieves an overall score of 63.2 compared with 68.9 for the teacher, retaining 91.7% of the teacher's performance while recovering 6.4 of the 12.1 points lost through pruning alone (56.8). Compositional reasoning performance remains close to the teacher on SugarCrepe (86.1 vs. 86.4), MR2-Bench (24.5 vs. 24.7), and ARO (59.5 vs. 60.4), while Winoground group performance (6.8 vs. 8.5) remains the primary limitation. Depth pruning also reduces forward latency by approximately 10% (150.0 to 136.4 ms per image on NVIDIA DGX Spark using FlashAttention-2). We present the architecture, compression methodology, training procedures, and evaluation results, demonstrating the effectiveness of Eddy-VL for multimodal retrieval under constrained edge deployment. Model weights and inference code are publicly available on Hugging Face.