Research map / Vision-language and multimodal / Human and embodied modeling
HieRD: Hierarchical Relational Distillation for Vision-Language Embedding Models
Vinh Le, Nguyen Hong Dang, Tu Vu, Linh Ngo Van, Duc Anh Nguyen, Trung Le
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| distill with HieRD instead of standard fine-tuning (FastVLM-0.5B) | FastVLM-0.5B student · MMEB CLS average | Precision@1: 64.93 -> 67.11 | Improved |
| distill with HieRD instead of standard fine-tuning (LLaVA-OneVision-0.5B) | LLaVA-OneVision-0.5B student · MMEB CLS average | Precision@1: 65.83 -> 70.08 | Improved |
| distill with HieRD instead of standard fine-tuning (LLaVA-OneVision-0.5B) | LLaVA-OneVision-0.5B student · DocVQA | Precision@1: 38.43 -> 49.23 | Improved |
| distill with HieRD on FastVLM-0.5B VQA | FastVLM-0.5B student · MMEB VQA benchmarks | Precision@1: avg 51.56 -> 52.76 | Mixed |
| add intra-modal structural alignment loss (ablation) | HieRD · SUN397 | Precision@1: 66.2 -> 67.2 | Improved |
| add cross-modal structural alignment loss (ablation) | HieRD · ImageNet-1K and VOC2007 | Precision@1: ImageNet 54.7 -> 56.0; VOC2007 78.9 -> 80.6 | Improved |
| add hidden representation alignment loss (ablation) | HieRD · ImageNet-1K | Precision@1: 55.7 -> 56.0 | Improved |
| use both intra-modal and cross-modal losses versus removing both (ablation) | HieRD · ImageNet-1K | Precision@1: 54.2 -> 56.0 | Improved |
| use attention-weighted group aggregation instead of mean pooling (ablation) | HieRD · SUN397 | Precision@1: 63.7 -> 67.2 | Improved |
| use attention-weighted group aggregation instead of mean pooling (ablation) | HieRD · ImageNet-1K | Precision@1: 53.7 -> 56.0 | Improved |
| change DBSCAN min_samples from 8 to 4 (ablation) | HieRD · MMEB CLS overall | Precision@1: 67.16 -> 66.12 | Hurt |
| change DBSCAN min_samples from 8 to 12 (ablation) | HieRD · MMEB CLS overall | Precision@1: 67.16 -> 66.58 | Hurt |
Closest papers by meaning
- VLM2Vec: Training Vision-Language Models for Massive Multimodal Embedding Tasks (ICLR 2025 · similarity 0.90)
- Bridging Vision and Language Spaces with Assignment Prediction (ICLR 2024 · similarity 0.89)
- Compositional Entailment Learning for Hyperbolic Vision-Language Models (ICLR 2025 · similarity 0.88)
- Towards General Continuous Memory for Vision-Language Models (NeurIPS 2025 · similarity 0.88)
- Remodeling Semantic Relationships in Vision-Language Fine-Tuning (AAAI 2026 · similarity 0.88)
- UVU: Improving Multimodal Understanding via Vision-Language Unified Autoregressive Paradigm (CVPR 2026 · similarity 0.88)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
