Research map / Architectures and efficiency / Pruning and efficient tuning
LoFA: Learning to Predict Personalized Prior for Fast Adaptation of Visual Generative Models
Yiming Hao, Mutian Xu, Chongjie Ye, Jie Qin, Shunlin Lu, Yipeng Qin, Xiaoguang Han
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 Stage-I LoRA response map prediction (ablation) | LoFA · Human action video (text) | FVD: 665.4 -> 589.8 | Improved |
| add Stage-I LoRA response map prediction (ablation) | LoFA · Identity image generation | Face Sim: 0.497 -> 0.548 | Improved |
| add Stage-I LoRA response map prediction (ablation) | LoFA · Text-to-video stylization | CSD-Score: 0.394 -> 0.427 | Improved |
| supervise Stage-I with response maps rather than a lightweight LoRA hypernetwork (ablation) | LoFA · Human action video (text) | FVD: 655.1 -> 589.8 | Improved |
| take base model weights as input with prompt via cross-attention instead of prompt as input (ablation) | LoFA · Human action video (text) | FVD: 653.7 -> 589.8 | Improved |
| use 2% response threshold instead of 1% (ablation) | LoFA · Human action video (text) | FVD: 634.7 -> 589.8 | Improved |
| use 2% response threshold instead of 3% (ablation) | LoFA · Human action video (text) | FVD: 622.5 -> 589.8 | Improved |
| inject Stage-I feature at 4th and 8th blocks (ablation) | LoFA · Human action video (text) | FVD: 660.1 -> 589.8 | Improved |
| attend Stage-I feature at more early layers (2 to 8) (ablation) | LoFA · Human action video (text) | FVD: 660.1 -> 680.9 | Hurt |
| increase training LoRA pairs | LoFA · Human action video (text) | quality | Improved |
| predict LoRA directly vs per-case LoRA training | LoFA · Human action video (text) | FVD: 609.5 -> 589.8 | Improved |
Closest papers by meaning
- C-LoRA: Continual Low-Rank Adaptation for Pre-trained Visual Models (NeurIPS 2026 · similarity 0.89)
- LoRAtorio: An intrinsic approach to LoRA Skill Composition (NeurIPS 2026 · similarity 0.89)
- LoRA Recycle: Unlocking Tuning-Free Few-Shot Adaptability in Visual Foundation Models by Recycling Pre-Tuned LoRAs (CVPR 2025 · similarity 0.88)
- Flat-LoRA: Low-Rank Adaptation over a Flat Loss Landscape (ICML 2025 · similarity 0.88)
- SDS-LoRA: Overcoming Anisotropic Gradient Scaling in Low-Rank Adaptation (NeurIPS 2026 · similarity 0.87)
- LoRA-X: Bridging Foundation Models with Training-Free Cross-Model Adaptation (ICLR 2025 · similarity 0.87)
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.
