Research map / Vision-language and multimodal / CLIP and zero-shot
Hierarchical Variational Test-Time Prompt Generation for Zero-Shot Generalization
Zhaoyang Wu, Fang Liu, Licheng Jiao, Shuo Li, Lingling Li, Xu Liu, Puhua Chen, Wenping Ma
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 |
|---|---|---|---|
| replace test-time prompt tuning with HyperTransformer prompt generation (ablation) | MaPLe · ImageNet OOD average (ablation, MaPLe-based) | accuracy: 62.31 -> 63.29 | Improved |
| add HyperTransformer prompt generation to MaPLe (ablation) | MaPLe · ImageNet OOD average (ablation, MaPLe-based) | accuracy: 60.28 -> 63.29 | Improved |
| add variational prompt generation to deterministic prompts (ablation) | HyperTransformer MaPLe · ImageNet OOD average (ablation, MaPLe-based) | accuracy: 63.29 -> 64.09 | Improved |
| add variational prompts on ImageNet-A (ablation) | HyperTransformer MaPLe · ImageNet-A | accuracy: 59.26 -> 60.71 | Improved |
| make variational prompts hierarchical (layer-conditioned) (ablation) | HyperTransformer MaPLe · ImageNet OOD average (ablation, MaPLe-based) | accuracy: 64.09 -> 65.89 | Improved |
| make variational prompts hierarchical on ImageNet-Sketch (ablation) | HyperTransformer MaPLe · ImageNet-Sketch | accuracy: 50.41 -> 52.71 | Improved |
| make variational prompts hierarchical on ImageNet-R (ablation) | HyperTransformer MaPLe · ImageNet-R | accuracy: 79.45 -> 81.45 | Improved |
| apply hierarchical variational prompt generation vs MaPLe+TPT | MaPLe · Cross-dataset average | accuracy: 66.50 -> 68.84 | Improved |
| apply hierarchical variational prompt generation vs strongest baseline DynaPrompt | DynaPrompt · ImageNet OOD average | accuracy: 64.31 -> 65.89 | Improved |
| apply hierarchical variational prompt generation vs TDA on EuroSAT | TDA · EuroSAT | accuracy: 58.00 -> 49.35 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- MaPLe — MaPLe: Multi-Modal Prompt Learning (CVPR 2023) · better on 3 of 3 shared results
- DynaPrompt — DynaPrompt: Dynamic Test-Time Prompt Tuning (ICLR 2025) · better on 2 of 2 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Black-Box Test-Time Prompt Tuning for Vision-Language Models (AAAI 2025 · similarity 0.90)
- On the Test-Time Zero-Shot Generalization of Vision-Language Models: Do We Really Need Prompt Learning? (CVPR 2024 · similarity 0.90)
- Any-Shift Prompting for Generalization over Distributions (CVPR 2024 · similarity 0.90)
- IPO: Interpretable Prompt Optimization for Vision-Language Models (NeurIPS 2024 · similarity 0.89)
- VaMP: Variational Multi-Modal Prompt Learning for Vision-Language Models (NeurIPS 2025 · similarity 0.89)
- DynaPrompt: Dynamic Test-Time Prompt Tuning (ICLR 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.
