Research map / Graph learning / Graph self-supervised learning
Beyond Graph Model: Reliable VLM Fine-Tuning via Random Graph Adapter
Bo Jiang, Xueyang Ze, Beibei Wang, Xixi Wang, Xixi Wan, Bin Luo
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 |
|---|---|---|---|
| add VRGAdapter to CLIP prototype baseline (ablation) | CLIP RN50 · ImageNet-1K (1-shot) | accuracy: 59.42 -> 62.49 | Improved |
| add VRGAdapter to CLIP prototype baseline (ablation) | CLIP RN50 · ImageNet-1K (16-shot) | accuracy: 64.46 -> 66.03 | Improved |
| add auxiliary models (MoCo, DINO) with averaging (ablation) | CLIP RN50 · ImageNet-1K (16-shot) | accuracy: 64.46 -> 69.84 | Improved |
| add auxiliary models (MoCo, DINO) with averaging (ablation) | CLIP RN50 · ImageNet-1K (1-shot) | accuracy: 59.42 -> 60.27 | Improved |
| add UMF fusion to auxiliary models (ablation) | CLIP RN50 + AUX · ImageNet-1K (1-shot) | accuracy: 60.27 -> 60.77 | Improved |
| add UMF fusion to auxiliary models (ablation) | CLIP RN50 + AUX · ImageNet-1K (16-shot) | accuracy: 69.84 -> 70.05 | Improved |
| add VRGAdapter to AUX+UMF model (ablation) | CLIP RN50 + AUX + UMF · ImageNet-1K (16-shot) | accuracy: 70.05 -> 71.35 | Improved |
| replace GraphAdapter with VRGAdapter (ablation) | CLIP+MoCo+DINO · DTD (1-shot) | accuracy: 54.08 -> 56.80 | Improved |
| replace GraphAdapter with VRGAdapter (ablation) | CLIP+MoCo+DINO · ImageNet-1K (16-shot) | accuracy: 70.83 -> 71.35 | Improved |
| use alpha = 0.7 instead of alpha = 0 (adapter-only features) (ablation) | VRGAdapter · ImageNet-1K (16-shot) | accuracy: 65.93 -> 71.35 | Improved |
| increase descriptions M from 2 up to 50 (ablation) | VRGAdapter · ImageNet-1K (16-shot) | accuracy: 71.35 | Improved |
| VRGAdapter instead of AMU-Tuning | RN50 · ImageNet-V2 (OOD) | accuracy: 58.64 -> 62.41 | Improved |
Closest papers by meaning
- p-Laplacian Adaptation for Generative Pre-trained Vision-Language Models (AAAI 2024 · similarity 0.91)
- MMA: Multi-Modal Adapter for Vision-Language Models (CVPR 2024 · similarity 0.88)
- GraphVLM: Benchmarking Vision Language Models for Multimodal Graph Learning (CVPR 2026 · similarity 0.88)
- Efficient and Context-Aware Label Propagation for Zero-/Few-Shot Training-Free Adaptation of Vision-Language Model (ICLR 2025 · similarity 0.88)
- Vision-Language Model Selection and Reuse for Downstream Adaptation (ICML 2025 · similarity 0.87)
- Exploiting the Asymmetric Uncertainty Structure of Pre-trained VLMs on the Unit Hypersphere (NeurIPS 2025 · similarity 0.86)
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.
