Research map / Architectures and efficiency / Pruning and efficient tuning
Parameter-Efficient Fine-Tuning for Pre-Trained Vision Models: A Survey and Benchmark
Yi Xin, Jianjiang Yang, Siqi Luo, Yuntao Du, Qi Qin, Haoxing Chen, Kangrui Cen, Yangfan He, Bin Fu, Yuewen Cao, Junjun He, Xiaokang Yang and 3 more
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 |
|---|---|---|---|
| use Adapter instead of full fine-tuning | ViT-B/16 · VTAB-1k | mean accuracy: 65.57 -> 71.44 | Improved |
| use HST instead of full fine-tuning | ViT-B/16 · VTAB-1k | mean accuracy: 65.57 -> 75.99 | Improved |
| use BitFit instead of full fine-tuning | ViT-B/16 · VTAB-1k | mean accuracy: 65.57 -> 62.05 | Hurt |
| use VPT-Shallow instead of full fine-tuning | ViT-B/16 · VTAB-1k | mean accuracy: 65.57 -> 64.85 | Hurt |
| use VPT-Deep instead of VPT-Shallow | VPT · VTAB-1k | mean accuracy: 64.85 -> 69.43 | Improved |
| use DTL for best performance-parameter trade-off | ViT-B/16 · VTAB-1k | PPT: 0.75 | Improved |
| fine-tune with AdaptFormer instead of full | VideoMAE ViT-B · SSv2 | top-1 accuracy (%): 53.97 % -> 59.02 % | Improved |
| freeze backbone instead of full fine-tuning | VideoMAE ViT-B · HMDB51 | top-1 accuracy (%): 46.41 % -> 49.84 % | Improved |
| use LoRA instead of full fine-tuning | Video Swin-B · SSv2 | top-1 accuracy (%): 50.99 % -> 38.34 % | Hurt |
| use BAPAT instead of full fine-tuning | Video Swin-B · SSv2 | top-1 accuracy (%): 50.99 % -> 53.36 % | Improved |
| use Prefix-tuning instead of AdaptFormer | Video Swin-B · HMDB51 | top-1 accuracy (%): 68.66 % -> 56.13 % | Hurt |
Closest papers by meaning
- Lessons and Insights from a Unifying Study of Parameter-Efficient Fine-Tuning (PEFT) in Visual Recognition (CVPR 2025 · similarity 0.95)
- Sensitivity-Aware Efficient Fine-Tuning via Compact Dynamic-Rank Adaptation (CVPR 2025 · similarity 0.91)
- Strong Baselines for Parameter-Efficient Few-Shot Fine-Tuning (AAAI 2024 · similarity 0.90)
- Parameter Efficient Fine-tuning via Cross Block Orchestration for Segment Anything Model (CVPR 2024 · similarity 0.90)
- Point-PEFT: Parameter-Efficient Fine-Tuning for 3D Pre-trained Models (AAAI 2024 · similarity 0.89)
- WST: Wavelet-Based Multi-scale Tuning for Visual Transfer Learning (AAAI 2025 · similarity 0.89)
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.
